Computerworld.com [Hacking News]

For years now, Microsoft has been doing its level best to move you from desktop Office and Windows to Microsoft 365, Windows 365, and Azure Virtual Desktop (AVD). Since the company first started down this road, however, something changed: the AI revolution, which has become a huge deal for the guys from Redmond. 

So, it should come as no surprise that the company is now combining its efforts to push cloud-based PCs and get as many users working with its AI services as possible.

First, Microsoft has reduced pricing for Windows 365 and AVD in select configurations by 20%. In particular, the company is slashing prices for persistent desktop deployments and lower-tier virtual machines (VMs). These are the instances commonly used by task workers and call centers, not by developers or white-collar office worker bees. 

Microsoft is also expanding bundled discounts tied to existing enterprise agreements and Microsoft 365 subscriptions. This cuts the per-user cost for organizations already invested in its Desktop-as-a-Service (DaaS) ecosystem. 

Beyond that, the company is introducing an on‑demand start experience that allows Cloud PCs to hibernate when users disconnect. The change will lower infrastructure costs while preserving performance when sessions resume. 

In addition to lower base pricing, Microsoft is promoting new autoscaling and power management features in AVD. For example, unused instances will now shut down automatically. This addresses one of the key criticisms of cloud desktops: the cost of idle capacity. 

Considering how PC prices are skyrocketing, this new DaaS pricing makes a lot of sense. But what the company giveth, the company taketh. Turns out that price cut comes as Microsoft 365 costs are set to rise in July. Microsoft wants you to think that a per desktop price increase of up to 33% is worth it for all the brand-spanking new AI goodness. Me? I’m still not sure Notepad Copilot improved my work life by 0.33%, never mind 33%! 

Windows prices are also going up. After July 1, standalone Windows components (per user/month, commercial list), such as Windows E3 and E5 are going up while Windows Enterprise (per device/month) is set to jump by 31%, from $5.85 to $7.63.

Overall, according to US Cloud, a leading independent provider of third-party Microsoft support and enterprise licensing optimization, Microsoft’s pricing decisions will mean a cumulative cost increase of up to 25% on a typical $10 million Enterprise Agreement by mid-2026. That is the company’s “AI Tax.” (Hey, those gigawatt AI centers aren’t going to build themselves!)

Meanwhile, back in DaaS land, Microsoft is positioning its AI‑enabled Windows 365 Cloud PCs as a higher‑end class of virtual desktops: Still in beta, they combine Windows 365 with AI acceleration to deliver integrated Windows AI features to any device via the cloud.

These high-end virtual AI desktops require Windows 365 Enterprise SKUs with at least 8 vCPU, 32GB RAM and 128GB storage. In other words, while the low-end virtual desktops are getting cheaper to use for AI, you’ll want the most expensive configurations. These currently cost $123 per user, per month.

By comparison, for serious AI use, Microsoft recommends Copilot+ PCs. These are high-end Windows 11 PCs powered by a turbocharged neural processing unit (NPU) capable of delivering more than 40 trillion operations per second (TOPS). These, such as Lenovo’s ThinkPad T14s Gen 6 Snapdragon, cost around $2,100. 

It might appear that the standalone PC is cheaper for AI. But Microsoft argues that its cloud PC — which you can run on any device — will scale well beyond what a fixed NPU can deliver in a Copilot+ PC configuration. 

What I see happening is Microsoft is using its lower-end DaaS price cut as a broadening of the funnel into Windows 365 while it keeps the AI‑heavy experiences as an upsell on more capable Cloud PCs. In other words, the low-end DaaS price cuts and the increases in PC prices are, Microsoft hopes, all about getting you to move to high-end AI cloud-based desktops. 

There’s also the rumor that Microsoft will be shifting its AI-cloud services from a fixed priced plan to a token-based one. True, for model APIs and gen‑AI services, the company had already standardized on token‑based pricing. But we always knew flat-pricing for AI services was a loss leader. If Microsoft can get you on a cloud-based AI desktop today, it’s betting it’ll get far more revenue from you tomorrow than it can with today’s Software-as-a-Service (SaaS) such as Copilot in Microsoft 365. 

If all this sounds complicated, you’re right — it is. As SAMexpert CEO Alexander Golev reports, “Microsoft AI uses two billing systems: Predictable per-user licensing and variable Azure consumption. Hybrid products like Copilot Studio and Security Copilot can charge through both, so costs land in different budgets, require separate monitoring, and complicate forecasting.” You think?

Of course, Microsoft has long been about shifting people from buying software to leasing it. Promoting Windows 365 AI‑enabled Cloud PCs is very much about using AI as the differentiator to justify running a Windows desktop from Azure instead of a cheaper, more generic VDI/DaaS stack or a more generic Windows DaaS. 

What all this boils down is that while Cloud PC list prices are coming down for small and midsize businesses, Microsoft is baking more AI into its Enterprise Cloud PCs. Thus, AI will be the value‑add that keeps the high end of its desktop cloud premium. This will all, Microsoft hopes, continue to empower growth, which increasingly relies on Azure and AI. 

So, what does this all mean for you and me? While I admit AI can be valuable in some areas such as programming, I remain cynical about AI in broader business uses. After all, as the MIT NANDA report, “GenAI Divide / State of AI in Business 2025,” states: “95% of organizations are getting zero return” from AI. 

Let’s say AI eventually does deliver the goods — or as I see it, we finally figure out how to get real value from the technology beyond just meeting summaries. That’s great. But it begs the question, “Do you really want to tie yourself to Microsoft’s desktop and service?” I don’t think so. 

Personally, I’ve never wanted to put all my desktop eggs in one Microsoft basket, and I’m sure not going to do that now with AI.  

Generative AI is fundamentally (and quickly) shaping how information is discovered and acted on, forcing enterprises to rethink how they engage with both humans and machines.

Adobe is responding to this shift, introducing new tools that keep up with evolving branding, surface campaign insights, and speed up content creation. At this week’s Adobe Summit, the company introduced a new Brand Intelligence system and expansions to its GenStudio platform, as well as launching products that use agentic AI to reshape customer experience.

And last week, Adobe announced one of its most significant AI rollouts, the new Firefly AI Assistant, interestingly on the same day rival Canva introduced its own new agentic platform, Canva 2.0.

“Brand visibility is going to shift toward what agents see, and what is picked up, understood, and returned by those agents,” said  Terra Higginson, a principal research director at Info-Tech Research Group. “Adobe is well-positioned for this shift.”

[ More Adobe Summit 2026 coverage ]

Helping teams stay on brand

Adobe describes Brand Intelligence as a “continuously-learning engine” that provides context to AI agents so teams can quickly create content that aligns with evolving branding. Instead of basic, static brand guidelines, the platform can gain from more nuanced insights like customer feedback, rejections, and approvals.

The system is based on a collection of small language models (SLMs) tailor-made to have a much deeper understanding of how a brand shows itself, in a multi-modal fashion, Adobe VP Sundeep Parsa explained to Computerworld. Enterprises can tweak the content along the way.

Brand Intelligence essentially reshapes the content supply chain so that teams can measure, plan, create, and manage brand-specific materials. Such customization is critical, because, he pointed out, “Coca Cola’s brand intelligence system is going to look very different from Nike’s.”

Info-Tech’s Higginson pointed out that an interesting use case for this would be when a brand wants to capitalize on a trend quickly. Back in the day, “instant” still meant a lot of planning and rework, but new Brand Intelligence capabilities can personalize agentic content immediately while still adhering to brand guidelines and localization requirements in real time.

“It’s taking personalized content to another level,” she said.

Ultimately, with LLMs in the mix, brand visibility is “twice as hard” as it used to be, and marketers’ jobs have gotten significantly more complex, Higginson noted.

“They don’t just need to convince humans to buy their products; they now need to convince agents, too,” she said. That means content must be instantly accessible, structured properly, and governed by brand guidelines.

New GenStudio enhancements

Along with Brand Intelligence, Adobe has rolled out new offerings within its GenStudio content creation platform. Notably, this includes a ‘Workflow Optimization Agent’ in Adobe Workfront, its workflow management platform.

The agent automates actions across workflows such as planning, execution, review, and approval. This can help teams build projects, speed up reviews, and pull out insights on demand, without the need for manual reporting. Teams can also involve AI agents in project planning and assign them tasks, such as resolving simple issues or performing reviews based on specific instructions and context.

Additionally, new creative production capabilities in Adobe Firefly for Enterprise Workflow Builder help developers build reusable workflows, link generative actions, and run batch production. Teams can also launch AI agents that interpret campaign briefs, compile supporting assets, and build templates and workflows.

Adobe is also rolling out a new dedicated ‘canvas’ interface to help teams pull together inputs and performance data to build better campaign briefs. Here, too, they can kick off an AI assistant.

Further, Adobe is announcing a new GenStudio module for marketing that turns long-form documents and videos into tailored campaigns, builds customer case studies, and writes web content. Marketers can quickly pull out performance insights (such as leads generated or follower statistics) and ask AI for recommendations to expand audience reach. And a new agency system of record will preserve enterprise context as it moves across content flows, supporting governance, accountability, and a shared understanding of branding.

“End users are seeing an exponential increase in the amount of content that needs to be created,” Higginson noted, adding that the benefits of a tool like Adobe GenStudio are not just generation of more content, but faster iteration, and less manual rework.

“It’s necessary because humans can basically no longer keep up with the demand for creation, iteration, testing, optimization, localization, and related workflows,” she emphasized.

Firefly AI Assistant speeds up creation

Adobe also announced a new Firefly AI Assistant (available soon) that users can interact with in natural language. Based on human instructions, agents can build workflows, ask contextual follow-up questions, offer suggestions, and report on results. They can also organize and share work in the Frame.io collaboration platform, interpret feedback, and automatically apply changes.

Agents have access to pre-built skills purpose-built for specific workflows, such as retouching portraits. They learn from creators over time, determining their preferred tools, workflows, and aesthetics, and maintain persistent memory across sessions. This carries over to other Adobe apps, too, so developers don’t have to start from scratch with each one.

A primary use case for the Firefly AI assistant is that of a creative team requesting input from the division owner, Higginson explained, including questions about color and brand consistency. In a demo at the event, Adobe illustrated this with a campaign for a major travel company. Feedback was folded in “almost in real time,” then localized across languages and countries.

“Work that would have taken months was reduced to minutes,” Higginson said. The platform makes creative workflows “more conversational and easier to manage across teams, which gets users to optimization much faster.”

Competition launches the same day

In an interesting development, Adobe announced its new Firefly AI Assistant on the same day that Canva introduced its next-gen agentic platform, Canva 2.0. Available now in research preview, Canva 2.0 also features brand intelligence tools and AI agents. Canva calls it its “most significant product evolution” since the company launched in 2013.

The platform upgrade reflects a broader move, beyond design “into more unified, AI-driven workflows,” Higginson said. Canva’s recent acquisitions of Simtheory and Ortto reinforce that push: They suggest the company is building toward a “wider marketing platform,” with content-centric workflow, stronger measurement, and lighter customer data platform (CDP) and lifecycle marketing capabilities that overlap with entry-level customer relationship management (CRM) territory.

That makes Canva more competitive, particularly with broader user groups and small to medium businesses (SMBs).

“Canva’s strength is accessibility, while Adobe’s strength is enterprise workflow depth,” said Higginson. “Adobe, though, still appears better positioned in complex enterprise environments where governance, scale, and established workflows matter more.”

LinkedIn is testing a new AI feature, Crosscheck, which allows users to compare several popular AI models directly on the platform. Users enter prompts into Crosscheck and receive two different responses generated by competing AI models from companies such as OpenAI, Anthropic, and Google.

After the user selects the best response, the model behind each answer is revealed. LinkedIn product manager Hari Srinivasan describes the service as a kind of blind taste test for AI models, according to Engadget.

Crosscheck works only with text, but has no limits on the number of questions. At the same time, LinkedIn shares anonymized user data with the AI companies to provide insights into how the models perform across different professional groups.

The feature is initially available to LinkedIn Premium subscribers in the US, with plans to expand to more countries and free users soon.

Adobe Summit serves as a platform for Adobe to introduce new services, capabilities, and enhancements to its portfolio of creative and marketing software and services. The 2026 edition kicks off live in Las Vegas on April 20, with a virtual event running alongside it.

The company has long been a name to watch as a developer of creativity and design software. This year will be a pivotal one for Adobe, with the recent announcement that its CEO is stepping down after 18 years at the helm, and the rise of AI tools challenging both creators and the software vendors that serve them. Expect announcements setting out Adobe’s response to those challenges — and watch for signs that the company may be changing direction or losing ground to AI offerings from companies such as Microsoft, OpenAI, or Google.

Adobe is also one of the bigger players in the enterprise digital marketing solutions, and Adobe Summit will also provide insights into changes to Adobe Marketing Cloud and Adobe Experience Platform.

Follow this page (and this one) for the latest news and insights from Adobe 2026, and check out recent related coverage below.

Adobe news and analysis Adobe bets on AI agents to stay at the center of marketing workflows

April 24, 2026: Facing pressure from rival design software vendors and general-purpose AI assistants, Adobe is rolling out agentic systems to coordinate and execute creative work across workflows and apps.

Adobe builds an ‘agentic content supply chain’ for the AI era

April 21, 2026: Adobe is adding new capabilities to GenStudio and Firefly to automate content creation and provide agentic assistants to keep up with branding as it evolves in the age of AI. The additions include a ‘Workflow Optimization Agent’ in Adobe Workfront, its workflow management platform.

Adobe bets on agentic AI to rewrite SaaS for customer experience

April 20, 2026: Adobe is shifting its approach to what it calls ‘Customer Experience Orchestration (CXO).’ Announced today at Adobe Summit, the new Adobe CX Enterprise suite is a pivot to a future defined by agents rather than by software alone, where SaaS companies claim an advantage based on their deep domain expertise and troves of first- and third-party data.

The top priority for Adobe’s next CEO? Prepping for the ‘age of agents’

April 9, 2026: Adobe CEO Shantanu Narayen announced plans to step down last month after 18 years leading software vendor. For whomever is tapped next for the top job — the CEO search is expected to take several months — the biggest priority will almost certainly be reshaping Adobe’s products and strategy for the next wave of agentic AI, analysts said.

Adobe Acrobat Standard review: A polished PDF editor with clear limits

March 16, 2026: Adobe Acrobat Standard is Adobe’s entry paid-tier PDF editor, sitting between the free Acrobat Reader and the more fully featured Acrobat Pro. It’s meant for people who need more than viewing, signing, and basic annotations, but not Adobe’s full set of advanced document tools. It handles routine PDF work well, but advanced features still require a move up to Pro.

Adobe CEO steps down after 18 years

March 13, 2026: It’s all change at the top for Adobe as CEO Shantanu Narayen is stepping down after 18 years. He will relinquish the role as soon as a successor has been found, although he will continue to serve as chairman. The creative software company’s new boss will have to face the challenge of AI.

Adobe makes Agent Orchestrator and AI agents generally available

September 10, 2025: Adobe has launched AEP Agent Orchestrator and six AI agents for building, delivering, and optimizing customer experiences and marketing campaigns. Agent Composer, coming soon, will aid with customizing and configuring agents.

Adobe Commerce and Magento users: Patch critical SessionReaper flaw now

September 10, 2025: Adobe issued an emergency patch for one of the most severe vulnerabilities ever discovered in the Magento Open Source ecommerce platform and Adobe Commerce, its enterprise counterpart. The flaw allows unauthenticated attackers to hijack user accounts and, in some cases, execute arbitrary code on servers. Security experts warn of exploits soon.

Adobe is developing an agent to integrate Adobe Express with Microsoft 365

March 20, 2025: Adobe is working with Microsoft to develop an AI agent that can generate graphics and design content from within the Microsoft 365 interface. The Adobe Express Agent, still under development, will allow users to create and embed graphics directly within productivity applications such as Word and PowerPoint.

Adobe makes agentic AI push with Agent Orchestrator, purpose-built agents

March 18, 2025: Adobe unveiled Agent Orchestrator, a new capability for Adobe Experience Platform (AEP) to supervise AI agents, whether from Adobe or third-party ecosystems. It also introduced 10 new agents built on the capability at Adobe Summit 2025.

Apple now has a new iCEO, as current leader Tim Cook (65) announced late Monday that he is set to become chairman of the board, while current head of hardware engineering, John Ternus, prepares to take over as CEO on Sept. 1.

As you’d expect, this leadership transition at one of the world’s most successful firms, is generating reams of news reports and hot takes. Here’s mine: Just as Steve Jobs presided over the resurrection of Apple and Cook led the company through unprecedented business growth, Ternus will guide the company through an era of equally unprecedented hardware proliferation. 

Expect more growth

He’s someone who cares about craft in hardware design and recently appeared in a worth-watching video interview (chaperoned by Greg Joswiak, the company’s senior vice president of worldwide marketing). The soon-to-be-CEO did well in what was an obvious media training exercise. “Everything we do, even if our customers don’t necessarily see it, everything we do has some new ideas in it…, we feel like we’re innovating all the time,” he said. 

Among a range of achievements, Cook innovated operations to the extent that every product Apple makes is supported by the world’s most efficient multinational manufacturing and logistics system. While he did, Ternus innovated product. “John Ternus has the mind of an engineer, the soul of an innovator, and the heart to lead with integrity and with honor,” said Cook.

The current CEO has had to handle huge supply chain challenges while scaling logistics to support growth. The numbers illustrate this: Apple sold 18.1 million Macs in the year following his appointment as CEO 15 years ago. In 2025, it sold 27.2 million. iPhone sales grew from 136 million in 2011 to 247 million last year. Rumor has it the MacBook Neo has shifted as many as 10 million units, just as Ternus turns up to take that figure higher.

“Tim’s unprecedented and outstanding leadership has transformed Apple into the world’s best company,” said outgoing Apple board chairman Arthur Levinson. 

You don’t need a weatherman

You can see which way the wind blows. 

Apple’s hardware is selling in record quantities, even as the company seems more prepared – and better able – than ever before to widen its addressable market with more affordable products. It is able to do this without compromising on product quality or user experience for three big reasons: 

• The massive per-customer services income built by Cook.
• Huge iPhone sales as an inheritance from Steve Jobs.
• The adoption of Apple Silicon, which has been presided over by Ternus and led by Johny Srouji.

Srouji will take on the hardware leadership role being vacated by Ternus and as part of this will combine the hardware technologies and hardware engineering teams, separated in 2012. “I am excited to bring these teams together and deepen their integration to help us innovate even more than we do today. There is no limit to what we can achieve together,” he wrote. 

That optimism is well-founded. Apple’s processor designs will enable the company to push fast in its new phase of proliferation. With 1nm chips on the horizon, Apple’s processors are small, powerful, and energy efficient, making them suitable for a plethora of new hardware designs the world hasn’t even seen yet.

Making impossible things possible

Apple under Ternus will no doubt lean into that opportunity. This approach means that not only will you see Apple widen its addressable market with a combination of product quality at better prices, but you’ll also watch it expand its offer with new product families. 

It’s no coincidence, for example, that Ternus at one point led Apple’s robotics team as the company prepares to introduce its first robotic products in the coming months. Apple’s hardware is supported by Apple’s software, of course. 

While it will offer some of its own solutions within Apple Intelligence, Apple doesn’t even need to make the AI. It just needs to make the best hardware to run AI on, which is what Ternus is going to focus on. 

You can already see it. With Ternus leading hardware, the Mac is more powerful and more popular today than at any point in its history — and the MacBook Neo is building on that success. It represents the thin end of a wider wedge of hardware-driven market share growth across all Apple’s products that will now accelerate under Ternus, even while the latter makes his own transition.

As board chairman, Cook will turn to handling the complex political and strategic relationships he’s been dealing with as CEO. Cook is very good at that, which also raises the question of whether he has wider ambitions for political engagement.

You can follow me on social media! Join me on BlueSky,  LinkedIn, Mastodon, and MeWe. 

Apple announced late Monday that Tim Cook, the company’s CEO since 2011, is stepping down Sept. 1 to be replaced by current senior vice president of hardware engineering, John Ternus. Cook will become executive chairman of the board.

Cook, who is 65, will continue as CEO until the end of August to assist in the transition, which, Apple said, came after a “thoughtful, long-term succession planning process.”

“It has been the greatest privilege of my life to be the CEO of Apple and to have been trusted to lead such an extraordinary company, Cook said In the announcement about the changes. “I love Apple with all of my being, and I am so grateful to have had the opportunity to work with a team of such ingenious, innovative, creative, and deeply caring people who have been unwavering in their dedication to enriching the lives of our customers and creating the best products and services in the world.”

John Ternus will succeed Tim Cook as Apple CEO in September.

Apple

He also sang his successor’s praises: “John Ternus has the mind of an engineer, the soul of an innovator, and the heart to lead with integrity and with honor. He is a visionary whose contributions to Apple over 25 years are already too numerous to count, and he is without question the right person to lead Apple into the future. I could not be more confident in his abilities and his character, and I look forward to working closely with him on this transition and in my new role as executive chairman.”

Current board non-executive chairman Arthur Levinson, who has held the post for 15 years, will become Apple’s lead independent director on Sept. 1. When he assumes his new role, Ternus will join the board.

Apple also announced that Ternus’s job will be filled by Johny Srouji, effective immediately.

“Johny is one of the most talented people I have ever had the privilege to work with,” Cook said about that change. “He has played a singular role in driving Apple’s silicon strategy, and his influence has been felt deeply not just inside the company, but across the industry. He has always led his organization with remarkable deftness and judgment, and time and again, his team has delivered breakthrough innovations that have transformed our products. We are incredibly fortunate to have him as Apple’s chief hardware officer.”

The ongoing shortage of memory chips looks likely to continue throughout the year as demand from the AI sector surges. According to Nikkei Asia, leading manufacturers are expected to be able to meet only about 60% of global demand despite expansion plans.

Although new factories are on the way, several of them are not expected to reach full production until 2027 at the earliest. Even once those facilities are up and running, additional time will be required to scale up to efficient production levels.

An annual production increase of around 12% would be needed to catch up with demand, analysts said, though current plans are significantly lower. The balance between supply and demand for memory is not expected to normalize until 2028.

Because of the shortage, memory prices have risen by approximately 90% during the first quarter of 2026.

Apple Silicon has another big journey to take, one that means Apple will probably be the first to introduce 1.4- and 1-nanometer chips inside its systems. If that happens, Macs, iPhones, and iPads will continue to lead the industry in performance per watt.

Why do I say this? Mainly because reports claim TSMC is working to build sub 1nm chips by 2029 — and Apple remains that company’s most important customer, despite competition from AI server manufacturers today.

Demand for AI servers could yet slow, given the looming energy crisis and the trend toward on-prem and edge AI services. I don’t think the current level of investment in AI is sustainable, which is why I think Apple will continue to be TSMC’s lead customer once that bubble, inevitably, bursts.

What’s happening at TSMC?

The latest news is that TSMC intends to begin trial production of its sub-1nm A10 process tech by 2029, setting up Apple to be the first big company to use these new processors inside its hardware when volume production begins. 

What’s interesting is that this move to 1nm isn’t just about making transistors smaller, but also about ensuring close integration between chips, memory, and energy systems. A report in 2021 said TSMC was able to reach 1nm by using bismuth instead of silicon in the design.

Apple, of course, already works very, very hard to integrate those different elements on its existing processors, which is why it delivers better performance at lower wattage than competitors. That integration means its systems can accomplish a great deal more from lower quantities of memory, which helps protect the company’s margins against rapidly accelerating RAM prices. 

We currently expect up to 30% improvement in both performance and power efficiency from these new chip designs. That implies that iPhone Pro models introduced in 2030 (or possibly 2031) will be powered by these new chips.

Apple’s silicon road map seems secure

TSMC is expected to introduce 1.6nm chips in the next 18 months, though Apple might choose to skip that iteration to guarantee a leadership position once the 1.4nm TSMC process hits in 2028. That iteration will deliver yet another big speed and performance boost to Apple’s devices, with Apple becoming the first PC, tablet, or smartphone manufacturer to ship 1.4nm systems at scale. 

What benefits can we expect? During TSMC’s 2025 North American Symposium the company said 1.4nm chips should be 15% faster and consume around 30% less power than the processors inside Apple’s current devices. That’s all good, but it is also interesting to note that the iPhone 17 series hasn’t even made the leap to 2nm as yet, with Apple using TSMC’s N3P process. So, the company has lots of scope to secure the future of Apple Silicon.

Where next for Apple’s chips?

If it is correct that Apple will skip TSMC’s 1.6nm process and then climb aboard the 1.4nm and 1nm chips, we could see the two big processor development chapters between now and 2030. This year we can see it introduce 2nm chips, with 1.4nm to follow probably in 2028 and the huge leap to sub-1nm processors to follow in 2030-31.

As these chips will be deployed across Apple’s hardware platforms, including within new designs we don’t know about yet, it means you can anticipate highly significant performance gains wherever in the ecosystem you happen to sit. Whether you’re looking at the next-generation MacBook Neo, MacBook Pro, iPhone or iPhone e, you’ll see impressive performance gains unlocked in all into the last half of this decade. 

Those performance gains, combined with improved energy consumption, allows Apple’s hardware designers to work towards thinner, lighter and smaller devices in a range of design configurations — some of which could not have existed before. (Think about spectacles with the kind of performance you once got from a Mac.) The way ahead is clear. Apple has a wide open road for chip design, and while tensions between today’s US and China could derail some of these plans, TSMC’s continued investment in fabrication capacity in the US might help mitigate against even that potential calamity.

You can follow me on social media! Join me on BlueSky,  LinkedIn, Mastodon, and MeWe. 

Enterprises have spent the past two years rushing to make their workforces “AI-ready.” But many early training programs — focused on prompt writing and chatbot skills — are proving poorly suited to the realities of AI-powered work.

The reason is simple: the skills that matter most once AI enters real workflows have less to do with interacting with tools and more to do with judgment. The durable capabilities emerging in the AI era include output validation, data literacy, process understanding, and the ability to challenge automated recommendations. Tool-specific skills, by contrast, tend to age quickly as models and interfaces evolve.

“AI-ready is not defined by how many people took training or how many licenses you bought,” said Neal Sample, executive vice president and chief digital and technology officer at electronics retailer Best Buy. “It’s defined by whether you have redesigned real workflows, assigned accountability, and can show the technology is improving outcomes without introducing unmanaged risk.”

That shift — from tool proficiency to operational judgment — is forcing enterprises to rethink how they train employees for AI.

The illusion of AI readiness

The first wave of corporate AI training focused heavily on prompt engineering and basic familiarity with generative AI tools. That approach made sense early on, when employees needed help understanding the technology. But many organizations are discovering those skills have a short half-life.

“Prompt engineering aged the fastest,” said Rebecca Schalber, senior manager for generative AI at cosmetics company cosnova Beauty. As new models and interfaces appear, the effort invested in crafting perfect prompts quickly becomes obsolete.

When cosnova rolled out generative AI across its workforce, Schalber expected training to center on individual capability — understanding large language models, learning prompting techniques, and experimenting with tools. Early adoption looked promising. Within six months, a survey showed employees reporting productivity gains of nearly 10%.

Adoption alone was not enough. “You need broad adoption to move the needle,” Schalber said. “But what really matters is the workflow design.”

Instead of focusing on prompts, cosnova began examining how work actually happens inside teams — what tasks employees perform, where friction exists, and which parts of a workflow could be safely automated or augmented by AI. That shift forced employees to confront a different question: not how to use AI, but how to verify its output and integrate it into real business processes.

When AI hits real workflows

The distinction becomes clear once AI leaves experimental environments and enters operational workflows. In testing, outputs can be compared against known answers. In real business processes, however, the answer often isn’t known in advance. AI systems are deployed precisely because they help employees analyze complex situations, interpret data, or generate insights.

That’s where human oversight becomes critical. “Human oversight is not second-guessing every output from the AI,” said Sample from Best Buy. “It means being explicit about where judgment, escalation, and accountability must remain human.”

The closer a decision comes to customer trust, regulatory obligations, or significant financial risk, the more important that judgment becomes. Organizations deploying AI at scale must build guardrails into workflows and clearly define who is responsible for final decisions.

“For every AI-enabled workflow, you need to know who owns the decision, who handles exceptions, and where a human must intervene before the business takes action,” Sample said.

In other words, the challenge of AI readiness is not teaching employees to interact with a model — it’s teaching them how to supervise it.

From training programs to workflow design

At cosnova, Schalber’s team moved away from generic training sessions toward hands-on workshops where managers and employees map their daily workflows. During these sessions, teams identify tasks that could benefit from AI support and then redesign processes around those opportunities.

When AI was introduced as simply another tool, enthusiasm was limited. But when employees saw how the technology could remove tedious tasks or reduce friction in their work, adoption accelerated.

“It was no longer just another tool that management wanted people to use,” Schalber said. Instead, teams were solving their own problems — removing repetitive tasks or speeding up processes they disliked.

The company also began emphasizing transferable skills that apply across AI tools and models, including critical thinking, workflow design, and data literacy. These capabilities remain valuable even as the technology evolves and have proven far more durable than prompt-writing techniques.

Experimentation before formal training

Some organizations are taking a different approach: encouraging experimentation first and formal training later. At AI infrastructure company Turing, Taylor Bradley, vice president of talent strategy, deliberately began the company’s AI upskilling effort by encouraging non-technical employees to experiment with generative AI tools.

The goal was to spark curiosity rather than enforce compliance. Bradley compares the process to teaching his daughter to ride a bicycle. “The best way for her to learn was to actually have her ride the bike,” he said.

At Turing, employees experimented with AI through informal activities such as turning photos of pets into “royal portraits” or creating short AI-generated films for internal competitions. The exercises were designed to lower the barrier to experimentation. Once employees became comfortable with the technology, the company introduced practical workshops focused on real work tasks.

Bradley now sits down with teams to examine daily workflows and identify where generative AI could help. Employees often discover that AI can serve as a sounding board for ideas, a drafting assistant, or a way to accelerate communication.

Within weeks, those experiments often evolve into more formal systems. One early project began as a conversational tool helping HR specialists draft responses to employee support tickets before expanding into a broader internal knowledge system.

The key metric, Bradley said, is not course completion but whether teams develop useful AI applications. “We focus on quality use cases with measurable outcomes,” he said.

Learning inside the flow of work

For large enterprises, the challenge of AI skill development is even more complex. Traditional training models — where employees attend courses and then return to their jobs — are poorly suited to technology evolving as quickly as generative AI.

According to Margaret Burke, talent acquisition and development leader at professional services firm PwC, traditional training programs are inherently episodic. “Employees attend a course, return to work, and may or may not apply what they learned,” she said. “In an AI-accelerating environment, that model breaks down.”

PwC is embedding AI learning directly into everyday work. The firm still runs formal programs but is expanding apprenticeship-style learning and weaving AI capability development into routine business activities.

One example is the company’s “skills days,” where employees explore AI applications relevant to their work. During a recent session with advisory associates, participants documented how they were already using AI — or where they planned to apply it. Hundreds of ideas emerged. PwC then used AI to analyze the inputs, clustering them into categories and redistributing the results across the organization so teams could learn from one another.

Crucially, PwC pairs technical AI capabilities with what Burke calls “human edge” skills, including critical thinking, independent judgment, and storytelling. “We never teach an AI technical skill without teaching the human skill that goes with it,” Burke said.

As AI systems generate more content and analysis, those human capabilities become essential for interpreting results, spotting errors, and explaining insights to colleagues and clients.

Measuring real AI readiness

As organizations rethink AI capability, the metrics used to evaluate training programs are changing. Traditional learning programs often rely on course completion rates or certifications. But those metrics reveal little about whether employees can use AI responsibly inside real workflows.

Instead, organizations are looking for operational signals. Some track how frequently employees develop new AI use cases that improve productivity or decision-making. Others measure how quickly teams adapt when AI tools or models change.

For Bradley at Turing, the key indicator is whether employees continually find new ways to improve their work with AI. “If my team members come to me every week with ideas for improving or expanding AI use cases, that’s the signal that capability is growing,” he said.

From the CIO perspective, however, the ultimate measure is operational outcomes. AI readiness only becomes meaningful when organizations integrate AI into real workflows while maintaining accountability for the results.

“The most durable capabilities are not the current best prompt tricks,” said Best Buy’s Sample. “They are judgment, problem framing, systems thinking, and the ability to translate machine output into business action.”

But for CIOs deploying AI across the enterprise, workforce capability is only part of the equation. Organizations must also rethink how leadership defines accountability when AI systems influence decisions.

“An AI-ready workforce without an AI-ready leadership model is likely to stall,” Sample said. “AI can accelerate analysis and recommendations, but accountability doesn’t transfer to the model. Leaders still have to define guardrails, decision rights, and what success looks like.”

As enterprises move beyond early AI experimentation, that leadership clarity may prove just as important as any skill employees learn.

Related reading:

• What AI skills job seekers need to develop in 2026
• 5 things IT managers get wrong about upskilling tech teams
• Two-thirds of jobs will be impacted by AI
• How to keep tech workers engaged in the age of AI
• How to train an AI-enabled workforce — and why you need to

Identity management is a critical concern for any enterprise, and it’s becoming ever more complex and convoluted with the advent of AI agents.

World ID is taking a unique (and to some, controversial) approach to this challenge by building a ‘digital proof of human’ ecosystem for the internet. Today, at its “Lift Off” event, the Sam Altman co-founded initiative made a series of announcements, which included the launch of version 4.0 of its World ID protocol, a World ID app, World ID for Business, World ID for Agents, a new verification tool called Selfie Check, new monetization programs, and integrations with Zoom and Okta.

“It’s a re-engineering of the stack around a very simple idea: Humans should have a right to exceptional privacy and security,” Daniel Shorr, chief of staff to the CEO at Tools for Humanity, said at the event.

How ‘proof of human’ works

Billed as the infrastructure for the age of AI, World ID was co-founded by Altman and Alex Blania, and is being developed by technology company Tools for Humanity, whose iris imaging technology seeks to eliminate the need to provide emails, photos, or other personal details to prove identity.

World ID’s mission is to provide “proof of human” (POH), so that people know they are in fact interacting with another human being (or a bot on behalf of a verified human), rather than a deepfake or other unknown entity. The ideal is to reduce abuse, impersonation, fraud, and misinformation, and promote trust in online interactions.

POH ensures that only one account exists per user (‘one-person-one-ID’) via Tools For Humanity’s iris-scanning Orb device, which uses multispectral sensors and infrared light to capture high-res images of a human’s irises. These images are processed in seconds on-device to generate an ‘IrisCode,’ a unique cryptographic hash based on the iris’s unique details and textures.

IrisCodes are then compared to entries in the World Chain, a global blockchain-based database, to verify the user hasn’t previously registered. This check uses zero-knowledge proofs (ZKPs), a cryptographic prover-verifier mechanism, to confirm iris uniqueness without needing to link personal data.

If the IrisCode is identified as unique, the user receives a World ID that can be stored on their phone. IrisCodes are anonymized and fragmented across secure servers to minimize breach risks, preventing reverse engineering. The Orb also deletes original images by default.

Other World ID initiatives include Deep Face and Face Auth, which help identify deepfakes by performing private 1:1 face comparisons of selfies and Orb-captured images.

Tiago Sada, chief product officer at Tools for Humanity, emphasized the protocol’s open source nature, third-party auditing, and regular security updates. “It goes beyond standard end-to-end encryption, and it uses multiple primitives, including anonymized multi-party computation and zero knowledge proofs to protect you along the way,” he said at today’s event.

More than 18 million people across 160 countries have now verified their “humanness” via Orb and have used them more than 450 million times, execs said.

New World ID features

The new World ID 4.0 is a more scalable and powerful version of World ID that incorporates essential upgrades like key rotation (which detaches keys from identity), multi-party entropy (to ensure that every interaction is unlinkable), and finer credential controls (more ways to manage and protect information), Shorr explained.

It now includes a new verification method, “Selfie Check,” that can be used in lieu of Tools for Humanity’s Orb device. “Take a selfie and ‘boom, you’re in,’” Shorr explained. He noted that it’s not as robust as the Orb, but it’s “really, really compelling for specific use cases. Not every use case today requires the gold standard of Orb assurance.”

World ID also now includes agent delegation tools that essentially serve as what Shorr called “a power of attorney for your agent,” allowing it to perform actions on the user’s behalf.

“With the explosion of agents, the internet is fundamentally changing again,” he said. “How do you make sure the right humans are in the loop?”

This is especially important at critical moments where users or platforms need to ensure that a purchase or decision was intentional. At the same time, he said, “we don’t want Skynet.”

Security company Okta is now onboard, introducing Human Principal, a verification method based on World ID that is now available in beta.

World ID also announced upcoming new monetization efforts. Shorr noted that it’s difficult to monetize the network when you can’t share user data, but at the same time, being human is “incredibly valuable” in the age of AI, and the internet will want to know which users are human.

“We dug through the history books, and we came up with an inventively old approach: Fees,” he said. When services or developers ask for World ID proof, apps will pay a fee, not humans.

World ID and Zoom fighting deepfakes

Ensuring participants in Zoom calls are real people is another concern.

Brendan Ittelson, Zoom’s chief ecosystem officer, noted that deepfakes are more realistic than ever and the technology to create them is much more accessible, so it’s no longer a hypothetical ‘will this happen?’

Customers across Zoom’s user base are deeply concerned, he said, yet there are challenges with existing verification techniques and knowledge base options.

“The technology is evolving so fast, so doing detection techniques and all that is a constant cat and mouse game,” he said. “You really need a platform where you’re looking at [the question], ‘how can you validate someone and be privacy forward, but also have that strong human connection?’”

To address that problem, today’s announcements included the news that World ID is coming to Zoom. New capabilities will match live images with the Orb-verified ID on a user’s device when they log into a call. They can also verify themselves in real time; nothing leaves their device. World ID verification will be indicated by a badge in the user’s Zoom window.

Not everyone is convinced, though

While touted as a way to make the internet a safer, more democratic, and inclusive place, the ambitious initiative has been met with significant criticism.

Detractors, including the likes of notorious whistleblower Edward Snowden, warn of privacy and biometric data risks. They argue that storing iris data could create immense security problems, as well as the potential for its misuse and for unlawful surveillance.

Other criticisms are that World ID creates a central point of failure, requires blind trust in one company, and exploits vulnerable and developing nations. For instance, the initiative became massively popular in Kenya because iris scans were traded for Worldcoin cryptocurrency (WLD). This hinted at bribery, detractors note; the program has since been banned in the country, and is also either banned or suspended in Brazil, Indonesia, Hong Kong, and Spain.

Further, the initiative raises concerns around data protection laws, credential theft (which can be particularly catastrophic because irises are immutable), and ‘function creep’ that could eventually restrict access to sites and force participation in the program.

Indeed, Orbs, which began shipping in the third quarter of 2025, are purchased from the private Tools for Humanity organization and are owned by “community operators,” who verify World IDs with their devices and receive WLD tokens for their efforts.

Protecting this kind of biometric data is crucial, said David Shipley of Beauceron Security: He pointed to Apple’s approach, where biometric data is securely stored on-device, and only a digital expression based on that data is transmitted, never the original biometric data itself.

“This feels like a super-bad idea,” he said of World ID. While having a secure, verified digital ID as a service that can be trusted is much needed, it shouldn’t be delivered by a private sector entity, he contended.

“Private sector control of personhood feels Hollywood-style cyber dystopian,” said Shipley. “Proof of being human and proof of being a citizen are public goods and should be delivered by public bodies that can be held accountable through democratic representation.”

Long before Taco Tuesday became part of the pop-culture vernacular, Tuesdays were synonymous with security — and for anyone in the tech world, they still are.  Patch Tuesday, as you most likely know, refers to the day each month when Microsoft releases security updates and patches for its software products — everything from Windows to Office to SQL Server, developer tools to browsers.

The practice, which happens on the second Tuesday of the month, was initiated to streamline the patch distribution process and make it easier for users and IT system administrators to manage updates.  Like tacos, Patch Tuesday is here to stay.

In a blog post celebrating the 20th anniversary of Patch Tuesday, the Microsoft Security Response Center wrote: “The concept of Patch Tuesday was conceived and implemented in 2003. Before this unified approach, our security updates were sporadic, posing significant challenges for IT professionals and organizations in deploying critical patches in a timely manner.”

Patch Tuesday will continue to be an “important part of our strategy to keep users secure,” Microsoft said, adding that it’s now an important part of the cybersecurity industry.  As a case in point, Adobe, among others, follows a similar patch cadence.

Patch Tuesday coverage has also long been a staple of Computerworld’s commitment to provide critical information to the IT industry. That’s why we’ve gathered together this collection of recent patches, a rolling list we’ll keep updated each month.

In case you missed a recent Patch Tuesday announcement, here are the latest six months of updates.

Microsoft’s Patch Tuesday release for April is a whopper

Windows admins are going to be busy this month, dealing with the largest Patch Tuesday cycle in memory. The April release involves 165 updates and roughly 340 unique CVEs from Microsoft — including two zero-days, one of which is already being actively exploited in the wild. 

The Readiness team is recommending “Patch Now” schedules for nearly every major product family this month: Windows, Office (with a zero-day), Microsoft Edge (Chromium), SQL Server, and Microsoft Developer Tools (.NET). April also brings Phase 2 of Microsoft’s Kerberos RC4 hardening with full enforcement set for July. There is a lot to cover, so the Readiness team built an infographic mapping the deployment risk for each platform.

More info is available here on Microsoft Security updates for April 2026.

For March, Patch Tuesday delivers fixes for 83 vulnerabilities

Microsoft’s March Patch Tuesday release addresses 83 vulnerabilities across Windows, Office, SQL Server, Azure, and .NET — with two publicly disclosed zero-days affecting SQL Server and .NET (though neither is being actively exploited in the wild.) Six additional vulnerabilities spanning the Windows Kernel, Graphics Component, SMB Server, Accessibility Infrastructure, and Winlogon are flagged as “Exploitation More Likely.”

The most significant change this month is the introduction of Common Log File System (CLFS) hardening with signature verification, which will affect how Windows handles log files across the operating system. More info on Microsoft Security updates for March 2026.

February’s Patch Tuesday release fixes 59 flaws, including 6 being exploited

The company’s Patch Tuesday release for February addresses 59 CVEs across the company’s product family — roughly half the volume of January’s 159 patches. Six vulnerabilities, affecting Windows Shell, MSHTML, Desktop Window Manager, Remote Desktop, Remote Access, and Microsoft Word, are already being actively exploited. (All five Critical-rated CVEs target Azureservices rather than Windows, however.) 

Both Windows and Office get a “Patch Now” recommendation, with CISA setting a March 3 enforcement deadline for all six exploited vulnerabilities. Two new enforcement timelines also take effect in April: Kerberos RC4 deprecation (CVE-2026-20833) and Windows Deployment Services hardening (CVE-2026-0386). More info on Microsoft Security updates for February 2026.

For January, Patch Tuesday starts off with a bang

The first Patch Tuesday release of 2026 addresses 112 CVEs across Microsoft’s product portfolio, including eight rated critical and three zero-day vulnerabilities. One zero-day (CVE-2026-20805), an information disclosure flaw in the Desktop Window Manager, is already under active exploitation, prompting CISA to add it to the Known Exploited Vulnerabilities catalog with a remediation deadline of Feb. 3, 2026. (Note: 95 of the vulnerabilities affect Windows.) More info on Microsoft Security updates for January 2026.

Ho ho ho! December’s Patch Tuesday delivers three zero-days

The December Patch Tuesday update addresses three zero-days (CVE-2025-64671, CVE-2025-54100, and CVE-2025-62221) but includes surprisingly few total patches (just 57). Notably, Microsoft has not published any critical updates for the Windows platform this month. That said, given the zero-days, we recommend a “Patch Now” release schedule for Windows and Microsoft Office. More info on Microsoft Security updates for December 2025.

Be thankful: November’s Patch Tuesday has just one zero-day

This November Patch Tuesday release offers a much reduced set of updates, with just 63 Microsoft patches and (only) one zero-day (CVE-2025-62215) affecting the Windows desktop platform. Windows desktops this month require a “Patch Now” plan, and while the severity of these security vulnerabilities is less than it was in October, the testing requirements are still extensive. More info on Microsoft Security updates for November 2025.

Windows admins are going to be busy this month, dealing with the largest Patch Tuesday cycle we can recall. The April release involves 165 updates and roughly 340 unique CVEs from Microsoft — including two zero-days, one of which is already being actively exploited in the wild. 

The Readiness team is recommending “Patch Now” schedules for nearly every major product family this month: Windows, Office (with a zero-day), Microsoft Edge (Chromium), SQL Server, and Microsoft Developer Tools (.NET). April also brings Phase 2 of Microsoft’s Kerberos RC4 hardening with full enforcement set for July. There is a lot to cover, so the Readiness team built an infographic mapping the deployment risk for each platform.

(More information about recent Patch Tuesday releases is available here.)

Known issues

Microsoft reports a single Windows 11 25H2 issue. It affects a narrow enterprise deployment group, but matters to anyone affected.

• KB5083769 – BitLocker recovery prompt on first restart (Windows 11 25H2/24H2). Devices with BitLocker enabled on the OS drive and the Group Policy “Configure TPM platform validation profile for native UEFI firmware configurations” set with PCR7 in the validation profile may be prompted for the BitLocker recovery key on the first restart after installing this update. Recommendation: Remove the PCR7 Group Policy configuration and run gpupdate /force before installing.

Issues resolved

April’s KB5083769 closes four issues, three quality-of-life and one multi-cycle reset failure:

• KB5083769 – Reset this PC (Windows 11 25H2/24H2). Resolves a defect that broke device reset on certain hardware and configuration combinations, taking the last-resort recovery path with it.
• KB5083769 – Secure Boot certificate rollout. The ongoing Secure Boot CA refresh picks up two improvements: the Windows Security app now displays certificate update status directly (Settings → Privacy & Security → Windows Security), and the quality update widens the device-targeting data for the staged rollout.
• KB5083769 – SMB compression over QUIC. SMB compression requests over QUIC now complete more consistently; the update addresses prior timeouts.
• KB5083769 – Remote Desktop anti-phishing. Opening a .RDP file now triggers a confirmation dialog listing every requested connection setting, each disabled by default. Users must explicitly opt in to local resource sharing before the connection is made; a one-time security warning appears the first time a .RDP file is opened after installing the update.

Major revisions and mitigations

Microsoft released no major revisions to Windows or Office. But Azure and Chromium/Edge have picked up several updates since the last month:

• Microsoft documented four critical Azure CVEs; no user action required.
• Microsoft re-published 141 Chrome/V8/WebGL/WebML/WebRTC fixes from the weekly upstream cadence; Edge picks them up through its own auto-update channel.

So Microsoft published 145 CVEs that affected Edge over the past 30 days. That averages out to around five reported security vulnerabilities per (working) day. Does anybody remember the good old days when we just had 10 critical-rated memory-related issues with IE — each month?

Windows lifecycle and enforcement updates

The saying that “April is the cruelest month” seems apropos, as we have three rather strict enforcements from Microsoft:

• Kernel driver cross-signed trust — evaluation mode begins April. Microsoft is dropping trust for legacy kernel drivers signed under the deprecated cross-signed root program, audit-only on Windows 11 24H2/25H2/26H1 and Server 2025.
• Kerberos RC4 hardening Phase 2 — April. Following November 2025’s Phase 1, domain controllers now default to AES-SHA1 encrypted tickets for accounts without an explicit Kerberos encryption type configured (CVE-2026-20833). The enforcement phase begins in July.
• Windows Deployment Services hands-free deployment — disabled by default from April. Hardening for CVE-2026-0386 (Unattend.xml over unauthenticated RPC) disables hands-free WDS deployment by default, beginning with the April update. Admins can override, but Microsoft does not recommend doing so.

Testing guidance

Each month, the Readiness team analyzes the latest Patch Tuesday updates and provides detailed, actionable testing guidance. April’s release covers 56 component updates across Windows. Microsoft flagged two as High Risk — Kerberos authentication and the Remote Desktop client — and delivered five patches to the Projected File System driver affecting cloud sync scenarios. Secure Boot and BitLocker validation expands to seven scenarios this cycle, including a new Windows Hello PIN persistence check. Prioritize Kerberos infrastructure, Remote Desktop stability, and cloud sync before broad deployment.

Kerberos and KDC

The Kerberos Key Distribution Center (kdcsvc.dll) and client library (kerb3961.dll) carry a High Risk flag this month. Microsoft’s guidance targets environments using keytab-based authentication with RC4 encryption — a legacy configuration common in mixed Windows and non-Windows service environments. The client-side update affects only Windows 10 1607, but server-side changes apply to all editions from Windows Server 2022 through 2025.

• After installing the update on domain controllers, open Event Viewer and review the System and Security logs for events with IDs 201–209.
• Capture full event details for any new events in that range: text, timestamp, and affected account or service.
• Focus testing on long-running services authenticating via RC4 keytabs, as these are most likely to surface failures after the update.

Remote Desktop client

Microsoft also flags the Remote Desktop ActiveX control (mstscax.dll) as High Risk. The update affects clipboard redirection, printer redirection, and session reconnection stability across all supported Windows versions. A separate update to mstsc.exe covers SmartScreen behavior for .RDP file handling, RemoteApp, and Hyper-V Enhanced Session mode.

• Connect to a remote device using mstsc.exe and check that the session establishes and remains stable.
• Copy and paste between local and remote sessions, both text and files, and expect correct transfer in both directions.
• Redirect a local printer into the remote session, print a test page, and confirm the job completes.
• Disconnect, reconnect, and verify clipboard and printer redirection survive the reconnection.
• Expect RemoteApp resources to launch normally and Hyper-V Enhanced Session mode to connect without error.

Secure Boot and BitLocker (continuing)

Secure Boot and BitLocker testing now expands to seven scenarios, including a new Windows Hello PIN persistence test. These validate Secure Boot state, BitLocker encryption, and key rolling related to the ongoing CVE-2023-24932 mitigation. Perform only on dedicated test devices with recovery keys backed up.

• Enable BitLocker on the OS drive, verify TPM protectors are present using manage-bde -protectors -get c:, then disable and verify the drive is fully decrypted.
• Enable BitLocker on a data drive, verify protectors, then disable and verify decryption completes.
• With Secure Boot enabled, enable BitLocker, trigger the recovery screen using reagentc /boottore, and verify the recovery key unlocks the drive.
• With Secure Boot disabled, enable BitLocker, force recovery via BCD test signing changes, unlock with recovery key, suspend BitLocker, and verify normal boot resumes.
• With both enabled, apply the Secure Boot key update (CVE-2023-24932) and verify the system boots without triggering recovery.
• Test hibernation with Secure Boot and BitLocker both enabled and verify clean resume without recovery prompts.
• On a device running March 2026, enable Windows Hello PIN and BitLocker, install the April update, and confirm the PIN still works.

Networking

April patches the Ancillary Function Driver for WinSock (afd.sys) twice — once paired with the TDX transport driver, once standalone — making it the most-patched network component this month. A separate patch to HTTP.sys affects HTTP/3 on Windows 11 23H2 and 22H2.

• Browse websites, download and upload files (including large files), and test VPN and Remote Desktop connections over both IPv4 and IPv6.
• Check that Teams, Outlook, and other messaging applications sign in, send messages, and reconnect after network blips.
• Test sandboxed and low-privilege processes — Edge, Store apps, and Electron apps — to confirm their network requests succeed.
• Generate sustained network load and confirm no BSODs, no new errors in Event Viewer, and no throughput degradation.

VPN and IPsec

April patches two VPN components: the Windows Filtering Platform driver (wfplwfs.sys) and the IKE Extensions service (ikeext.dll). The WFP update targets UWP VPN plug-in stability, sleep/wake recovery, and Always On VPN. The IKE update covers IKEv2 tunnels, IPsec security associations, and Connection Security Rules.

• Connect and disconnect your UWP VPN plug-in client repeatedly (10+ cycles) and confirm the client remains usable and the system stays stable.
• Keep the VPN connected for 30+ minutes during active use; verify it survives network changes (Wi-Fi to Ethernet) and sleep/wake cycles.
• If using Always On VPN, confirm it connects at sign-in and reconnects after network loss.
• Establish IKEv2 VPN connections and verify the tunnel is stable and internal resources are reachable.
• Validate that Connection Security Rules negotiate IPsec correctly and that protected traffic remains protected.

Authentication and security

Patches to the SSPI kernel drivers (ksecdd.sys, ksecpkg.sys) span NTLM, Kerberos, CredSSP, and TLS/SSL. The Windows Hello for Business stack also picks up updates for Enhanced Sign-in Security.

• Exercise end-to-end sign-in and resource-access flows for applications that use NTLM, Kerberos, CredSSP, or TLS/SSL authentication.
• Test both success and failure cases: correct versus incorrect credentials, allowed versus denied accounts, and expired certificates.
• Verify Windows Hello for Business authentication with Enhanced Sign-in Security across sign-in, lock, unlock, and reboot cycles.

Graphics, Shell and desktop

April updates span Direct3D, the Desktop Window Manager, and the graphics kernel (win32kbase.sys, win32kfull.sys). The Windows Shell (shell32.dll) picks up a patch affecting Mark-of-the-Web preservation for downloaded shortcuts, and COM Automation (oleaut32.dll) gets an update.

• Run stress tests with sustained UI activity: rapid open/close of windows, snap layouts, virtual desktop switching, and multi-monitor connect/disconnect.
• Test GPU-accelerated workloads — video playback, 3D applications, browser hardware acceleration — and check for visual artifacts or flickering.
• Download a .lnk shortcut file from the internet and confirm SmartScreen displays a warning when the shortcut is opened — verifying Mark-of-the-Web is preserved.
• Run COM Automation workflows — VBA, PowerShell, and Office automation — and confirm they execute correctly.

Hyper-V and virtualization

April patches both Hyper-V compute layers (computecore.dll, vmcompute.dll, vmwp.exe), along with the hypervisor binary (hvax64.exe) for Windows 11 25H2 and 24H2.

• Start, save, resume, and stop a VM using Hyper-V Manager or PowerShell and repeat the cycle multiple times.
• Export a VM, import it, and confirm the imported VM boots and runs normally.
• Launch Windows Sandbox and confirm it starts without error.

Windows Installer, Cloud Sync and MDM

April updates to Windows Installer (msi.dll), the Cloud Files filter (cldflt.sys), and the MDM management layer affect installation workflows, cloud sync, and device management.

• Install, uninstall, and repair MSI packages to verify Windows Installer functions correctly.
• Connect and disconnect your cloud sync provider (e.g. OneDrive) multiple times and confirm sync functions after restarts.
• Enroll a device in Intune or your MDM solution, verify compliance status, and trigger a policy sync.

Common Log File System and storage

The Common Log File System driver (clfs.sys) — subject of March’s major hardening change — picks up a follow-up patch. Storage Spaces (spaceport.sys) and app isolation file system drivers (bfs.sys, wcifs.sys) also receive updates this cycle.

• Run Windows Update install and rollback cycles, then power-cycle the machine multiple times to confirm the system boots normally each time.
• Install and uninstall a set of representative applications through multiple cycles and confirm each completes without error.
• Perform a backup using your normal solution, restore from it, and verify data integrity.
• If using Storage Spaces, create a pool with mirrored and thin virtual disks, write data, and verify clean deletion.

Office and SharePoint

April’s Office updates target MSI editions: Excel 2016 (KB5002860), PowerPoint 2016 (KB5002808), Office 2016 shared libraries (KB5002859), and SharePoint Server 2016, 2019, and Subscription editions. These will not install on Click-to-Run deployments such as Microsoft 365 Apps.

• Open and edit complex Excel workbooks with formulas, macros, and external data connections; save and reopen to verify integrity.
• Create and edit PowerPoint presentations with embedded media and transitions.
• Across all patched server editions, validate SharePoint document library operations, co-authoring, and workflow execution.
• Verify that Office add-ins and line-of-business applications integrating with Office continue to operate correctly.

April’s two High Risk components should top every testing queue. Kerberos changes could disrupt long-running services using RC4 keytabs; monitor event IDs 201–209 and keep rollback plans ready. The Remote Desktop client update warrants thorough validation of clipboard, printer redirection, and session reconnection, particularly in RDP-dependent environments. Secure Boot and BitLocker validation remains essential as CVE-2023-24932 key rolling continues. Five patches to the Projected File System driver elevate cloud sync testing this cycle. The dual afd.sys updates and VPN/IPsec patches warrant regression testing across remote-access infrastructure. Office updates are confined to MSI editions.

Each month, we break down the update cycle into product families (as defined by Microsoft) with the following basic groupings:

• Browsers (Microsoft IE and Edge)
• Microsoft Windows (both desktop and server)
• Microsoft Office
• Microsoft Exchange and SQL Server
• Microsoft Developer Tools (Visual Studio and .NET)
• Adobe (if you get this far)

Browsers

Microsoft’s browsers look quiet this month. Two Microsoft-authored Edge spoofing fixes both ride the standard Edge update channel: CVE-2026-33119 (Edge for Android, CVSS 5.4, moderate) and CVE-2026-33118 (CVSS 4.3, low).

The real story is upstream: 140+ Chromium fixes in the past month, including CVE-2026-5281 — a use-after-free in Dawn that Google has confirmed is actively exploited in the wild. We  recommend you patch now for all Chromium endpoints (here’s looking at you, Edge).

Microsoft Windows

Microsoft delivers 134 Windows CVEs across desktop and server — four critical, the rest important or moderate, with no zero-days or publicly disclosed flaws this cycle. Headline by raw CVSS is a 9.8 IKE/IPsec RCE; priority by exploitability is the Active Directory RCE — the only Windows critical Microsoft rates “Exploitation More Likely.” The four critical-rated issues are concentrated in three Windows areas: Active Directory, networking (two flaws), and Remote Desktop Client.

• Active Directory / Identity — CVE-2026-33826, RCE in Active Directory via improper input validation (CVSS 8.0, critical; Exploitation More Likely). An authenticated low-privilege attacker on an adjacent network can execute code on a domain controller – your entire directory service is the surface. This is a priority for anyone running AD on-prem.
• Networking (IKE/IPsec) — CVE-2026-33824, RCE in IKE Service Extensions via double-free (CVSS 9.8, critical; Less Likely). Highest CVSS in the cycle: unauthenticated, network-callable, no UI. Patch VPN concentrators and IPsec gateways first.
• Networking (TCP/IP) — CVE-2026-33827, RCE via race condition in the TCP/IP stack (CVSS 8.1, critical; Less Likely). Network-callable, but the race lifts attack complexity (AC:H).
• Remote Desktop Client — CVE-2026-32157, RCE via use-after-free (CVSS 8.8, critical; Less Likely). Triggered when a user connects to a malicious RDP server (UI:R) — the threat model is reverse RDP, not inbound. Flag for jump-host operators.

Beyond the criticals, the standout Windows flaw is CVE-2026-27912 — Kerberos elevation of privilege via improper authorization (CVSS 8.0, important). Authorized attackers on an adjacent network can elevate through the Kerberos handler. Coordinate domain-controller deployment with the Kerberos RC4 Phase 2 hardening covered in the lifecycle section; both touch domain controllers. The Kerberos flaw (CVE-2026-27912) pushes April’s Windows updates to Patch Now.

Microsoft Office

Office receives 14 security fixes, three rated critical and one actively exploited in the wild. The active SharePoint exploit forces Office to Patch Now, with SharePoint servers taking priority over the client push.

• CVE-2026-32201 – Microsoft SharePoint Server — Spoofing, actively exploited in the wild (CVSS 6.5, important). The score understates the urgency: exploitation has been confirmed, and a spoofing flaw inside SharePoint is a platform for credential theft and lateral movement regardless of internal-only posture. Patch immediately, ahead of the Office client push.
• CVE-2026-32190 – Microsoft Office — Remote code execution (CVSS 8.4, critical). The Preview Pane remains the attack vector; previewing a crafted file in Outlook or File Explorer is sufficient to trigger execution without further user action. As we’ve noted before, this keeps recurring.
• CVE-2026-33114, CVE-2026-33115 — Microsoft Word — Remote code execution (both CVSS 8.4, critical). Paired Word RCEs on the same release channel; affected surface matches CVE-2026-32190.

Excel carries the heaviest cluster — four additional RCEs: CVE-2026-32189, CVE-2026-32197, CVE-2026-32198, and CVE-2026-32199, plus an information-disclosure flaw in CVE-2026-32188. Microsoft Word picks up two fixes outside the critical pair: RCEs CVE-2026-33095 and CVE-2026-23657, and information disclosure CVE-2026-33822. This is a Patch Now release for Office, driven by the SharePoint zero-day. Organizations that cannot deploy Office clients quickly should consider disabling the Preview Pane in Outlook and File Explorer as a temporary mitigation against the critical RCE trio.

Microsoft Exchange and SQL Server

Exchange Server picks up zero CVEs this month, a rare quiet cycle, and the right window to clear any deferred CU work. SQL Server gets three, including a network RCE that grants SQL sysadmin on success:

• CVE-2026-33120 — Microsoft SQL Server — Remote code execution via untrusted pointer dereference (CVSS 8.8, important; Exploitation Less Likely). Authenticated attackers get full SQL sysadmin on success. Scope is unusually narrow: only SQL Server 2022 for x64-based Systems on the GDR servicing branch — CU 24 and every other supported version (2016 SP3 through 2025) are not listed as affected.
• CVE-2026-32167, CVE-2026-32176 — Microsoft SQL Server — Elevation of privilege via SQL injection (both CVSS 6.7, important). Paired flaws affecting SQL Server 2016 SP3 through 2025 on both GDR and CU branches. Local EoP, not remote — the concern is breadth, not blast radius.

The Readiness team recommends Patch Now for any SQL Server 2022 GDR operation. Schedule the wider SQL footprint with your normal database-maintenance window.

Developer tools

There are 10 CVEs in Developer Tools this month, headlined by a critical-rated .NET Framework DoS and two GitHub-attributed flaws that will affect developer workflows directly.

• CVE-2026-23666 — .NET Framework — Denial of service via improper input validation (CVSS 7.5, critical; Exploitation Less Likely). The critical rating despite a DoS impact reflects exploit-code maturity; the CVSS vector includes E:P (proof-of-concept).
• CVE-2026-32631 — Visual Studio — NTLM hash leak via git clone from manipulated repositories (CVSS 7.4, important). GitHub-attributed: cloning a malicious repo or checking out a branch that resolves to an attacker-controlled UNC path leaks the user’s NTLM hash. Affects Visual Studio 2017, 2019, and 2022 (17.12 and 17.14).
• CVE-2026-26143 — PowerShell — Security feature bypass (CVSS 7.8, important). Highest CVSS in the set, and PowerShell SFBs always merit attention.

Five more developer-related updates round out the cycle: four .NET / Visual Studio DoS or spoofing fixes (CVE-2026-26171, CVE-2026-32178, CVE-2026-32203, CVE-2026-32226) and a moderate TLS PSK/ALPN bypass (CVE-2026-21637). None have been disclosed or exploited. The Readiness team recommends Patch Now for .NET Framework and PowerShell.

Adobe (and third-party updates)

Microsoft no longer ships Adobe updates as part of its bulletin. Adobe ships APSB26-44 separately for Acrobat and Reader — two listed as critical. They are worth your attention, given Reader’s prevalence on enterprise desktops. For anyone packaging, testing and deploying these recent and rapid Adobe releases: we hear you. The packages are big, and the management effort keeps growing.

Has IPv6 finally reached its day of glory?

It’s fair to say that IPv6 has not had the level of take-up expected when the Internet Engineering Task Force (IETF) ratified it back in 1998. Take-up has been agonizingly slow, not reaching 5 percent of traffic until 2014. However, the use of IPv6 has been slowly climbing since, and according to Google statistics, briefly accounted for 50.1% of the internet traffic Google sees on March 28.

However, technology publication The Register, which spotted the tiny but significant blip in Google’s traffic graphic, quoted two other sources: Cloudflare and APNIC Labs as stating that IPv6 had yet to reach such an exalted level: Cloudflare tracked it at a high of 43 percent, while APNIC registered that 43.13% of network hosts across the world were IPv6 capable.

It has been a long climb to this point. IPv6, with support for around 3.4 x 1038  addresses, was developed due to fears that the 4.3 billion unique addresses available under the previous version of the protocol, IPv4, would be insufficient for a global population now numbering around 8 billion.

While the development of technologies such as Network Address Translation has extended the lifespan of IPv4 by allowing multiple devices to hide behind a single address, there is little doubt that IPv6 has gradually been growing in importance and there is every chance that the 50 percent usage line will be crossed for good at some point in the future.

This article first appeared on Network World.

Mark Zuckerberg, the CEO of Meta, is building an AI version of himself.

The virtual CEO is being trained on Zuckerberg’s mannerisms and will be loaded with his views on corporate strategy, the Financial Times reported.

The idea is that employees will find the virtual Zuckerberg more accessible than they would the flesh and blood manifestation.

There are plenty of claims that AI will lead to jobs being eliminated but, until now, the CEO job has looked safe. If Zuckerberg’s experiment proves successful, though, even company leaders could be due for the chop.

In February, OpenAI’s Sam Altman warned that CEOs could be as vulnerable as other senior executives. “AI superintelligence at some point on its development curve would be capable of doing a better job being the CEO of a major company than any executive, certainly me,” Altman said. “On our current trajectory, we believe we may be only a couple of years away from early versions of true superintelligence.”

Klarna CEO Sebastian Siemiatkowski has already tempted fate, using an AI version of himself to present the company’s financial results to analysts, and even to take customer calls. So far, though, he’s kept his job.

The UK government has created a Sovereign AI investment fund with up to £500 million (US$675 million) to spend on turning UK startups into national AI champions.

Its support could involve investments of up to £20 million per startup, or provision of up to 1 million GPU-hours of AI compute, and fast-tracking of visas to bring skilled workers to the UK.

The multi-million-pound budget sounds impressive, but it’s just 0.08% of OpenAI’s recent $852 billion valuation. That company just received fresh investment of $122 billion, dwarfing the UK’s sovereign fund.

Closer to home, that £500 million would buy about 5% of French AI startup Mistral, which has achieved its success by offering a European alternative for businesses that do not want to use American or Chinese AI providers.

The UK government does not have a great record when it comes to investing in national IT champions. In the 1960s and 1970s, the government ran the National Enterprise Board which provided funding to new technology companies, but even the biggest names helped in this way have slipped out of UK ownership: ICL, a mainframe challenger to IBM, eventually became part of Japan’s Fujitsu, while Inmos, an early innovator in parallel computing, is now part of Dutch chip giant STMicroelectronics.

Apple Business is aimed at small businesses coalesced around Macs, iPhones, and iPads. If that’s you, and all your systems are made by Apple, the service is likely to be all you need to run a small operation of up to a few dozen seats. 

But Apple Business isn’t really designed to handle the advanced needs of larger enterprises. And while it can provide a starting point for Mac deployments in mixed-platform environments, it probably shouldn’t be where you end up.

It doesn’t handle cross-platform device deployments, for which you’ll need full-strength MDM solutions (such as those from up-and-coming vendor Fleet). Another thing Apple Business doesn’t do is cover the full extent of compliance targets you might need to meet at your company. So, if you need to ensure compliance with standards/benchmarks such as HIPAA, SOC 2, ISO 27001, or CIS, you’ll need to choose something else.

This is also true if you need to ensure your endpoints are secured, or you require automated vulnerability scanning. 

A gift to small enterprise

That’s not to say Apple Business doesn’t have its uses. It clearly does. If you run a small business with up to, say, 50 staffers and you use Apple kit across the company, you’ll be able to manage your devices and app deployments yourself, no admin required.

That makes it a great tool for high-growth startups, many of which use Apple right from the start. Those businesses will be able to manage devices across their teams for free using Apple Business. They can always scale up once business is booming, making the service a gateway to tech success for many startups or small enterprises. The ability to streamline device management company-wide at no charge is a gift.

Setting the stage

Many might feel that with the international introduction of Apple Business, the company has torn a chunk out of the MDM industry. That’s less true than it sounds; many in the space already support small deployments for free, so what Apple is doing is winnowing away some of the smaller businesses who might use the resources provided by MDM firms but never become paying customers. 

Those customers are also an excellent market for the AppleCare support the company offers alongside Apple Business. It gives people the experience of device management, so that by the time they shift to a more advanced plan to support growth, they have a better understanding of what that involves.

Apple has drawn a line in the sand with the business. It’s basically saying that on the SMB side of that strip, it has you covered — and it has effectively defined its rapidly maturing MDM partners as focused on the needs of large customer deployments.

Market opportunity knocks

The good news there is that those large deployments do actually exist. In the last three years, Apple has confirmed huge Mac deployments (thousands of Macs) at SAP, Snowflake, Capital One, Coppel, Nubank, and elsewhere. Just last year, Apple CFO Kevan Parekh confirmed the best ever June quarter for Mac in the enterprise, and with the MacBook Neo, the company seems to be seeing dramatic growth in every one of the 200 markets in which you can now sign up for Apple Business.

So, while Apple nurtures tomorrow’s big businesses, its MDM partners can continue to meet the more diverse and demanding needs of larger enterprise entities. 

With the low-cost Neo arguably emerging to be the company’s iPhone moment for the Mac, Apple is also building business fast in emerging markets. Since use of Apple Business remains an integral component of working with any third-party device management partner (if only to assign the devices to an MDM system), the opportunity exits to scale up for business growth and scale down if that market contracts. It’s a world-class, ecosystem-based set of functionalities to support small business and enable corporations, all in one place.

You just need to know which problems it solves. Deployment? Yes. Compliance, edge security, and cross-platform support? No.

You can follow me on social media! Join me on BlueSky,  LinkedIn, and Mastodon.

No matter what type of Android phone you carry or how you usually use it, one thing is a near-universal constant:

You’re gonna spend a ton of time messing with messages.

The messages may be from clients, colleagues, or your cousin Crissy from Cleveland (damn it, Crissy!). But regardless of who sends ’em or what they’re about, they’re all poppin’ up on your phone and cluttering your weary brainspace.

My fellow Android adorer, I’m here to tell you there’s a better way.

Google’s Android Messages app has gotten surprisingly good over the years. That’s no big secret. If you only rely on what you see on the surface, though, you’re missing out on some of Messages’ most powerful and underappreciated efficiency-enhancing options.

[Hey: Want even more advanced Android knowledge? Check out my free Android Shortcut Supercourse to learn tons of time-saving tricks — for messaging and beyond!]

Today, we’ll explore the Android Messages app’s most effective out-of-sight superpowers. They may not be able to cut down on the number of messages you send and receive on your phone (DAMN IT, CRISSY!), but they will help you spend less time fussing with ’em. And they might just help you have a more pleasant experience, too.

Let’s dive in, shall we?

(Before you splash forward, take note: The tips on this page are all specific to the Google Messages app for Android. If you’re using a phone where that exact app wasn’t preinstalled or set as the default, you can download it from the Play Store and give it a whirl. You might be pleasantly surprised by what you find.)

Android Messages trick #1: Message resurrection

We’ll start with a freshly added fix for one of the longest-standing Android messaging frustrations — and that’s the app’s inability to let you bring back a messaging thread you inadvertently deleted.

As of just this month — April 2026 — Google is finally in the midst of adding an overdue “Trash” section into the Messages app that lets you see and optionally restore any conversation you’ve killed within the past 30 days. Can I get a halle-frickin’-lujah?!

Once the feature is available to you, it couldn’t be much easier to find and manage:

• Just tap your profile picture in the Messages app’s upper-right corner.
• And look for the newly added “Trash” option in the menu that appears — directly beneath “Archived.”

At last! A place for finding and optionally restoring deleted messages within the Google Messages app on Android.

JR Raphael, Foundry

Tap that, and you’ll be taken to a special section of the app where every axed thread lives for about a month before being permanently deleted. There, you can tap any thread and then tap the circular clock icon at the top of the screen to restore it — or tap the trash icon to delete it permanently (and irreversibly) from your phone. You can also restore or delete all threads in that area together, if you want.

Thank goodness.

Android Messages trick #2: Text editing

In addition to recovering a deleted message, Google Messages on Android also now allows you to edit your own sent message — for those mortifying moments when you realize you’ve made a mistake or sent the wrong thing to the wrong person (egad!).

The option is available only in messages where everyone involved is using the current RCS messaging standard. You’ll know that’s the case if you see “RCS message” in the text box at the bottom of the thread.

Provided that’s present, just press and hold your finger onto any individual message you’ve sent, then look for the pencil-shaped editing icon along the app’s upper edge — and, last but not least, swear to yourself you’ll never hit send without reading over what you wrote again. (Right…)

On that note, I’d be remiss not to inform you that anyone’s original, pre-edited messages are always accessible for anyone else in the conversation to see — if they know where to look.

Android Messages trick #3: Custom icons

Up next is what might be my favorite little-known trick within Google’s Android Messages app: With a couple quick adjustments, you can turn any of your contacts’ faces into a custom notification icon. That icon will then show up at the top of your phone whenever that person messages you for extra-easy visibility and access.

See?

A quick bit of simple setup, and bam: Anyone’s face can become their notification icon (for better or for worse!) on your phone.

JR Raphael, IDG

The only catch is that your phone needs to be running 2020’s Android 11 operating system or higher for the feature to be available. (And honestly, if your phone isn’t running Android 11 at this point, you’ve got bigger fish to fry, Francesco.) Also, Samsung has screwed around with this system for no apparent reason — a frustratingly common theme with Samsung’s heavily modified approach to Android, especially as of late — so you may or may not be able to take advantage of this on a Galaxy gadget, depending on how recently its software has been screwed up updated. (Exaggerated sigh. What more can I say?!)

On any reasonably recent Android device that sticks close to Google’s core Android interface, though, here’s how to make the magic happen:

• The next time you get a message from someone, press and hold your finger to the notification.
• That’ll pull up a screen that looks a little somethin’ like this:

Android’s Priority conversation setting is the key to creating custom notifications that really stand out.

JR Raphael, IDG

• Tap the “Priority” line, then tap “Apply” to save the changes.

And that’s it: The next time that person messages you, you’ll see their profile picture in place of the standard Messages icon in your status bar, and the notification will show up in a special section above any other alerts.

Hip, hip, hoorah!

Android Messages trick #4: Custom sounds

In addition to making it easier to spot an important contact by their notification icon, you can also create a custom alert sound for messages coming in from different people — or even from specific threads within the Google Messages app — so you immediately know what they are before you have a chance to look.

This is one of those things that’s super-basic but also awkwardly out of sight and consequently unknown to an awful lot of Android-owning organisms. But once you know where to find it, it really couldn’t be much easier to get going. And it’s all connected to Android’s notification channels, which let you get incredibly nuanced on how different types of notifications within apps behave.

The quickest way to zip where you need to be is to open the thread you want to customize within Messages itself — whether it’s a one-on-one text with an individual person or a group conversation with multiple contacts. Once you’re inside the thread, tap the three-dot menu icon in its upper-right corner and select “Details,” then select “Notifications” on the screen that comes up next.

And hey, wouldya look at that? You should now be staring at a series of options about how that exact notification behaves — including, at least in the standard Google version of Android, the all-important “Sound” setting.

width="1024" height="923" sizes="auto, (max-width: 1024px) 100vw, 1024px">A special sound for every conversation is no more than few fast taps away.

JR Raphael, Foundry

Tap that, then find and select any sound you like. The next time a new message comes in for this conversation, there’ll be no mistaking what awaits you from the second it arrives.

Android Messages trick #5: Contact prioritizing

While we’re thinking about making certain conversations stand out, ever wish you could keep your most important messaging threads at the top of the list for easy ongoing access?

Poof: Wish granted. No matter what kind of Android phone you’re holding or how needlessly meddled with its software may be, just hold your finger onto the conversation in question on the main Messages app screen, then tap the pushpin-shaped icon in the app’s upper bar.

You can now pin up to 20 conversations that way, and they’ll always appear above all other threads in that main inbox view.

Android Messages trick #6: Short-term tune-out

Some incoming messages always require your immediate attention. Others, in contrast, are best tuned out and caught up with much later.

Surely you’ve been in that type of thread before, right? Y’know, the one where people are getting just a little too active — sending message after message, typically either during your workday or while you’re trying to focus on anything other than their uninteresting missives?

Google’s Android Messages app actually has a great way to deal with such struggles. It’s a temporary snoozing mechanism that lets you opt out of notifications from one specific conversation and that conversation only and stop receiving alerts from it for a set amount of time.

To try it out, press and hold your finger onto any thread in your main Messages list, then look for the clock icon with a “z” inside of it at the top of the screen. Tap that, and you can then decide to snooze notifications from that single conversation for one hour, eight hours, 24 hours, or — if you really want to tune it out — forever.

The sanity-saving Android Messages snooze option. (Don’t worry: I won’t tell.)

JR Raphael, Foundry

You’ll continue to get notifications from all other conversations in the meantime, and you’ll still be able to see new activity from the snoozed thread by opening up Messages and actively looking at it. But you won’t be interrupted by its alerts again — for a little while, at least.

Android Messages trick #7: Gemini be gone

Speaking of tuning out distractions, if you rarely to never interact with Google’s Gemini chatbot from inside the Messages app — perhaps because, y’know, it’s also available in approximately 7 gazillion other in-your-face places — you might appreciate the distraction-free satisfaction of an interface without a prominent Gemini button begging for you to caress it every frickin’ time you open up your messages.

(The Gemin icon is that starbust-shaped symbol that sits perpetually above the “Start chat” button, in Messages’ lower-right corner, if you haven’t ever tapped it to find out.)

And good news: You can actually send that icon a-packin’, if you’re so inspired: Just tap your profile picture in Messages’ upper-right corner, select “Messages settings,” then tap “Gemini in Messages” and flip the switch on the screen that comes up next into the off and inactive position.

All that’s left is to pat yourself on the back and celebrate the fact that you’ll only have to see Gemini in 6.9 gazillion other places from this point forward.

Android Messages trick #8: Free in-flight Wi-Fi

Gemini may be little more than a distraction within Messages most of the time, but one moment when that capability can actually come in handy is while you’re flying the allegedly friendly skies.

A while back, a crafty reader alerted me to the fact that you could use Gemini’s Messages integration to effectively give yourself free in-flight Wi-Fi access even when you haven’t paid for the privilege. It’s a heck of a clever hack and a moment when you genuinely may want to activate Gemini within the Messages app, at least for a little while.

I’ve got a complete breakdown of how it works and how you can put it to use on your next flight.

Android Messages trick #9: Calendar connection

Whether from the sky or on the regular ol’ ground, the next time you’re working to plan a meeting or event with a fellow Homo sapien in Messages, make yourself a mental note of this:

Anytime someone sends you a message that includes a specific date and time, the Messages app will underline that text. See it?

That underlined time is a covert link from an incoming message to your Android calendar agenda.

JR Raphael, IDG

You’d be forgiven for failing to realize, but you can actually tap that underlined text to reveal a shortcut for opening that very same day and time in your Android calendar app of choice. It’s a great way to get a quick ‘n’ easy glimpse at your availability for the time you’re discussing.

And if you then want to create a calendar event, just look for the “Create event” command that should appear right below that very same message. That’ll fire up a new calendar event for you on the spot, with the appropriate day and time already filled in.

That button to the left of the text suggestions is a spectacular time-saver for on-the-fly event creation.

JR Raphael, IDG

Don’tcha just love simple step-savers?

Android Messages trick #10: Seamless scheduling

If you’re ready to hammer out a response to a message right now but don’t want your reply to be sent for a while, follow the advice shared by a reader in my Android Intelligence newsletter recently and simply schedule your message for some specific future time.

The Android Messages app’s scheduling system is spectacularly useful. You can rely on it for setting reminders to be sent to clients, business-related messages to be pushed out the next morning, or context-free middle-finger emojis to be delivered to your cousin in Cleveland at ungodly hours in the middle of the night.

To tap into this productivity-boosting power, just type out your message normally — but then, instead of tapping the triangle-shaped send icon at the right of the composing window, press and hold your finger onto that same button when you’re done.

No reasonably sane person would possibly realize it, but that’ll pull up a hidden menu for selecting precisely when your message should be sent.

Send any message, anytime — no matter when you actually write it.

JR Raphael, IDG

And the person on the other end will have no way of even knowing you wrote the thing in advance.

Android Messages trick #11: Swift saving

When you run into a message you know you’ll want to reference again, save yourself the trouble of trying to dig it back up later and instead star it on the spot to make it fast as can be to find in the future.

It couldn’t be much easier to do: Whilst viewing an individual message thread, just press and hold your finger onto the specific message you want to save, then tap the star-shaped icon that appears in the bar at the top of the screen.

Then, when you want to find the message again, tap the search icon at the top of the main Messages screen and select “Starred” from the menu that comes up. That’ll show you every message you’ve starred for exceptionally effortless resurfacing.

Android Messages trick #12: Smart searching

Speaking of that Messages search system: Starring is sublime, but sometimes, you need to dig up an old message that you didn’t go out of your way to save.

The Android Messages app makes that even easier than you might realize. Tap that same search icon at the top of the app’s main screen — and in addition to searching your entire history message for any specific string of text, take note:

• You can start typing out the name of anyone in your contacts, then select them from the suggestion that appears — and then type in some text to look for something specific only within messages from that one person.
• You can use the options within the main Messages search screen to look specifically at images, videos, locations, or links people have sent you.
• And you can combine any of those variables for even more granular finding — looking for links you sent to a particular client, for instance, or locations an out-of-town colleague sent to you.

The Android Messages app’s search system is chock-full of helpful info.

JR Raphael, IDG

How ’bout them apples?!

Android Messages trick #13: Instant marking

I don’t know about you, but I find it impossibly irksome to see messages sitting with bold emphasis in my Android Messages inbox. That, to me, is a marking that means I need to read (and possibly also respond) to the message in question. And I can’t possibly rest for the day until I know that everything in my Messages inbox is open, addressed, and dealt with (or at least opened and with a reminder set to deal with it at some specific future time).

Sometimes, though, it’s all too easy to fall behind and get a backlog of bolded messages — and in such scenarios, sometimes, you need a quick ‘n’ easy one-switch reset button to bring everything back to read status and give yourself a fresh start.

Well, surprise: Messages has such an option! Tap your profile picture in the app’s upper-right corner and look for “Mark all as read” in the menu that comes up to find it — then let yourself rest easy as all that attention-demanding boldness melts away once and for all.

Android Messages trick #14: Easier-to-read text

On the subject of more noticeable text, file this next Android Messages feature under “accidental discoveries”: The next time you find yourself squinting at something in a messaging thread on your phone, try a good old-fashioned zoom gesture on the screen — placing your finger and thumb together and then spreading ’em slowly apart.

You’d never know it, but the Messages app supports that standard gesture for zooming into a conversation. The inverse applies, too: When you’re ready to zoom back out and make everything smaller, just bring your two fingers closer together.

And if those actions aren’t working for you, tap your profile picture in the upper-right corner of the main Messages screen and select “Messages settings,” then make sure the toggle next to “Pinch to zoom conversation text” is in the on position.

Android Messages trick #15: Custom colors

While we’re thinkin’ about easier reading, a relatively recent Android Messages addition can let you create a custom color palette for any conversations you’ve got goin’.

That way, you can always remember that texts with your significant other are in, say, purple, whereas messages with your most important client are in red. (Best not to get those two threads confused.)

This one works only with messages sent using the modern RCS messaging platform, which basically means messages that don’t involve pesky people still carrying around iPhones with outdated software on ’em. (It’s always the iPhone people, isn’t it?!)

With any currently supported conversation, open up the thread within Messages — then:

• Tap the three-dot menu icon in the screen’s upper-right corner.
• Select “Change colors” from the menu that appears.
• Pick the color scheme you prefer, then tap the Confirm button at the bottom.

Every Android Messages conversation can have its own distinctive color, if you take the time to set it up.

JR Raphael, IDG

Repeat for any other compatible conversations, and you’ll always know exactly what you’re looking at even with a fast glance — and without having to give it an ounce of active thought.

Android Messages trick #16: Meatier media

You know a fantastic way to waste time? I’ll tell ya: moving from one app to another just to glance at something someone sent you (like those blasted Bangles video Crissy is always blasting your way).

But get this: Google’s Android Messages app can let you preview and get the gist of both text articles and even YouTube videos without ever leaving your current conversation — from right within the app and that very same message thread.

The key is to make sure you’ve got the associated options enabled:

• Tap your profile picture in the upper-right corner of the main Messages screen.
• Select “Messages settings,” then tap “Automatic previews.”
• Make sure the toggle next to “Show all previews” is on and active.

Now, the next time someone sends you a link, you’ll see the associated item’s thumbnail and description right then and there, within the Messages conversation:

Videos expanded in-line within Messages — easy peasy.

JR Raphael, IDG

With web pages, Messages will show you just enough of a preview to let you make an educated decision about whether you want to tap the link or not.

Web links gain useful extra context once you enable the right option within the Android Messages settings.

JR Raphael, IDG

Almost painfully sensible, no?

Android Messages trick #17: Smarter shortcuts

If I had to pick the simplest Android Messages trick for enhancing your efficiency, it’d be embracing the built-in shortcuts Google gives us for faster message actions.

From the main Messages screen, you can swipe left or right on any message to perform an instant action — archiving the conversation, deleting it, or toggling it between read and unread status.

All you’ve gotta do is mosey your way back into the Messages app’s settings areas and tap on the “Swipe actions” item to set things up the way you want…

Step-saving swipes within Messages — now available for your customization.

JR Raphael, IDG

…and then, just remember to actually use those gestures moving forward. (That part’s on you.)

Android Messages trick #18: Quicker cleanup

Certain services love to send confirmation codes via text messaging when you sign in or try to perform some action. It may not be the most advisable or effective form of extra security, but — well, it’s better than nothing. And for better or for worse, it’s a pretty common tactic.

Core security considerations aside, the most irksome part of these confirmation codes is having ’em clutter up your messages list at every Goog-forsaken moment. But the Google-made Android Messages app can actually take care of that for you, without any ongoing effort — if you take about 20 seconds to make the right tweak now.

Here’s the secret:

• Tappity-tap that comely character in the upper-right corner of the main Messages screen (y’know, the one whose appearance has a striking resemblance to your oversized head).
• Tap “Messages settings” in the menu that comes up, then select “Messages organization.”
• Within that curiously created section, you’ll see only one option: “Auto-delete OTPs after 24 hrs.” OTP may not exactly be an everyday, universally known abbreviation, but fear not — for it isn’t an erroneous reference to an early 90s rap hit with equally ambiguous meaning. Nope: It stands for one-time password, which is the same thing we’re thinking about here.
• Flip that toggle into the on and active position, then flip a finger of your choice to all the confirmation codes in your messages list with the knowledge that they’ll be auto-purged a day after their arrival from that point forward.

Who’s down with OTP? Every last homie. (I apologize.)

Android Messages trick #19: Readable reactions

Slack-style reactions may seem silly on the surface, but they serve an important communication purpose in allowing you to quickly acknowledge a message without having to carry the conversation on further. Whether it’s a thumbs-up, a clapping hands symbol, or even perhaps an occasional burrito emoji, it really can be a handy way to say “Yup, got it” (or “Yup, want beefy goodness”) without having to use a single word.

You probably know you can summon a reaction within the Android Messages app by pressing and holding a specific message within a conversation and then selecting from the list of available emoji options — right? But beyond that, Messages packs an even faster way to issue a reaction in the blink of an eye.

And here it is: Simply double-tap your finger onto any individual message within a conversation. That’ll apply the thumbs-up reaction to it without the need for any long-press or symbol selection.

It’d be nice if there were a way to customize which reaction is used for that action by default — so that, obviously, we could all change it to the burrito emoji, since that’s what any sane person uses most often — but if and when an upward thumb will do the job, now you’ve got a super-easy way to bring it into any conversation with a fast finger tap.

Android Messages trick #20: iRritation elimination

Last but not least in our list of magnificent Messages enhancements is something specific for your conversations with the Apple-adoring animals in your life. And it relates to those very same sorts of reactions we were just going over.

One obnoxious side effect of Apple’s “no one exists outside of iOS” mentality, y’see, is the way the iPhone’s equivalent of those reactions show up on Android. Plain and simple, they show up as — well, plain and simple text messages, instead of coming through as reactions.

Surely you’ve encountered this, right? Those pointless messages you get from iGoobers that say stuff like “Loved ‘Please stop texting me, Crissy'”?

Well, scribble this on your metaphorical mental iPad: Google’s Android Messages app is actually able to intercept those absurd platform-specific reactions and turn ’em into standard reactions instead of plain-text interruptions. And it’ll take you all of 12 seconds to enable the option:

• Head back into the Messages app’s settings.
• Tap “Advanced.”
• Look for the line labeled “Show iPhone reactions as emoji” and make sure the toggle next to it is in the on position.

All that’s left is to breathe one final heavy sigh of relief — and to send Crissy a well-deserved burrito reaction.

Hey: Don’t let the learning stop here. Get six full days of advanced shortcut knowledge with my free Android Shortcut Supercourse. Tons of time-saving tricks await!

The productivity gains from AI are so great, companies can lay off thousands of employees and still get the same amount of work done — right? Or maybe it’s the opposite: despite all the hype, any supposed AI productivity boom is a mirage, causing employees, even  developers, to experience heavier workloads.

At the moment, the jury’s still out on whether AI use boosts or busts productivity across the workforce, despite the prediction that American business spending on AI will exceed $200 billion by the end of the year, according to one analysis. 

There’s no doubt workers are turning to AI in a variety of ways. Gallup, for instance, says nearly half of all US workers now use AI. And Hubstaff data published by Worklytics shows that 85% of professionals use the technology— but only for about 4% of their actual work time. That means 96% of work is 100% human. 

Mileage varies according to how you group employee types, too. The Federal Reserve Bank of St. Louis found that workers using AI saved 5.4% of their work hours, a 1.1% overall increase in productivity. That’s an average, with math and computer workers and within the information services industry reporting higher productivity gains. 

The Federal Reserve Bank of Atlanta, meanwhile, uncovered what it called a “productivity paradox,” in which the productivity gains people think they see aren’t reflected in measurable gains. (It sounds like AI isn’t the only one hallucinating.)

And research from Harvard Business Review (February 2026) found that AI often increases the intensity of work rather than reducing the total workload as originally promised. I’ve heard software developers, in particular, expressing this view and finding that AI is a major source of job burnout. 

All this talk about productivity can miss the qualitative dimension. A 2025 study found that using AI makes employees more innovative by giving them confidence they can handle more complex problems. 

The research goes on and on and, taken together, is more or less inconclusive. However, it’s reasonable to assume that productivity gains from any kind of new technology are likely to take time to show up. It took a decade or more with the PC revolution, for example. While these early days for AI present a mixed picture, productivity gains will surely come, and probably on a massive scale. 

Meanwhile, one slice of the American workforce is already seeing giant gains — remote workers. 

Why AI is working for those working from home

As I’ve argued in this space many times, remote work is a boon for companies in most circumstances. The reasons for this bullish stance are both numerous and, to me, intuitive to the point of being obvious. 

Here are three: 

• Employees have more time because they don’t waste time commuting
• Flex hours are more likely with remote work, so employees can better manage work-life balance, making them happier and more committed to their jobs
• Remote work reduces interruptions, facilitating “deep work,” which, according to deep work expert Cal Newport, is the more valuable type of work for companies 

Now, a new study has added another major benefit for companies in allowing employees to work remotely: AI. 

The study by Michael Blank, a faculty fellow at the Stanford Institute for Economic Policy Research (SIEPR), and colleagues found that AI has a much higher impact in the home than in the office. The study looked at internet browsing data of more than 200,000 U.S. households. 

One reason is surprising: AI helps work-from-home (WFH) employees with both professional and personal tasks, making them more productive at both. The study shows that AI helps people save time and complete tasks much more efficiently when working, planning travel, shopping, figuring out how to fix things around the house and more. 

WFH employees have an AI advantage over office workers, according to the study, because they have the autonomy to integrate AI into their flow without corporate oversight and control. 

Also: Remote employees are more likely to task-switch during the day, alternating between work and personal tasks, something AI facilitates through increased automation. 

Interestingly, the researchers found that employees are taking time saved and using it for more leisure time, as opposed to doing more work or learning new skills. This particular fact is a mixed bag for employers, because while they’re not realizing productivity gains in terms of work performed, they are benefiting from happier employees less prone to dissatisfaction and burnout. 

Blank’s major note of caution is that he found younger people with higher incomes saw the highest productivity gains with AI use at home. He fears a growing “digital divide” between higher and lower income groups and younger and older workers.

It’s about the autonomy as much as the technology

I want to be very clear about the great revelation of this study. It does not look directly at higher productivity with the use of AI for work tasks. Nor does it necessarily conclude that only WFH remote employees can see these gains. 

What it found is that people with high autonomy are the ones who see  the biggest productivity gains from the use of AI in general. WFH employees have the highest autonomy, so they’re seeing real improvements in increased leisure time. 

Just as the benefits of “flex work” are not about flexibility in location but in the use of time, flexibility in the use of AI drives productivity. 

I’ve been beating the flex work drum for years, and now during the AI revolution I’d like to add autonomy to that mix. Whether employees are working in offices full or part-time, from home full or part-time or as digital nomads full or part-time, in 2026 it appears that the highest productivity and employee satisfaction gains come from maximizing flex work and AI autonomy. 

AI disclosure: I don’t use AI for writing. The words you see here are mine. I do use a variety of AI tools via Kagi Assistant (disclosure: my son works at Kagi) — backed up by both Kagi Search, Google Search, as well as phone calls to research and fact-check. I use a word processing application called Lex, which has AI tools, and after writing use Lex’s grammar checking tools to find typos and errors and suggest word changes. Here’s why I disclose my AI use and encourage you to do the same.

Anthropic has today released a new, improved Claude model, Opus 4.7, but has deliberately built it to be less capable than the highly-anticipated Claude Mythos.

Anthropic calls Opus 4.7 a “notable improvement” over Opus 4.6, offering advanced software engineering capabilities and improved visioning, memory, instruction-following, and financial analysis.

However, the yet-to-be-released (and inadvertently leaked) Mythos seems to overshadow the Opus 4.7 release. Interestingly, Anthropic itself is downplaying Opus 4.7 to an extent, calling it “not as advanced” and “less broadly capable” than the Claude Mythos Preview.

The Opus upgrade also comes on the heels of the launch of Project Glasswing, Anthropic’s security initiative that uses Claude Mythos Preview to identify and fix cybersecurity vulnerabilities.

“For once in technological history, a product is being released with a marketing message that is focused more on what it does not do than on what it does,” said technology analyst Carmi Levy. “Anthropic’s messaging makes it clear that Opus 4.7 is a safer model, with capabilities that are deliberately dialed down compared to Mythos.”

‘Not fully ideal’ in some safety scenarios

Anthropic touts Opus 4.7’s “substantially better” instruction-following compared to Opus 4.6, its ability to handle complex, long-running tasks, and the “precise attention” it pays to instructions. Users report that they’re able to hand off their “hardest coding work” to the model, whose memory is better than that of prior versions. It can remember notes across long, multi-session work and apply them to new tasks, thus requiring less up-front context.

Opus 4.7 has 3x more vision capabilities than prior models, Anthropic said, accepting high-resolution images of up to 2,576 pixels. This allows the model to support multimodal tasks requiring fine visual detail, such as computer-use agents analyzing dense screenshots or extracting data from complex diagrams.

Further, the company reported that Opus 4.7 is a more effective financial analyst, producing “rigorous analyses and models” and more professional presentations.

Opus 4.7 is relatively on par with its predecessor in safety, Anthropic said, showing low rates of concerning behavior such as “deception, sycophancy, and cooperation with misuse.” However, the company pointed out, while it improves in areas like honesty and resistance to malicious prompt injection, it is “modestly weaker” than Opus 4.6 elsewhere, such as in responding to harmful prompts, and is “not fully ideal in its behavior.”

Opus 4.7 comes amidst intense anticipation of the release of Claude Mythos, a general-purpose frontier model that Anthropic calls the “best-aligned” of all the models it has trained. Interestingly, in its release blog today, the company revealed that Mythos Preview scored better than Opus 4.7 on a few major benchmarks, in some cases by more than ten percentage points.

The Mythos Preview boasted higher scores on SWE-Bench Pro and SWE-Bench Verified (agentic coding); Humanity’s Last Exam (multidisciplinary reasoning); and agentic search (BrowseComp), while the two had relatively the same scores for agentic computer use, graduate-level reasoning, and visual reasoning.

Opus 4.7 is available in all Claude products and in its API, as well as in Amazon Bedrock, Google Cloud’s Vertex AI, and Microsoft Foundry. Pricing remains the same as Opus 4.6: $5 per million input tokens, and $25 per million output tokens.

What sets Opus 4.7 apart

Claude Opus is being branded in the industry as a “practical frontier” model, and represents Anthropic’s “most capable intelligent and multifaceted automation model,” said Yaz Palanichamy, senior advisory analyst at Info-Tech Research Group. Its core use cases include complex coding, deep research, and comprehensive agentic workflows.

The model’s core product differentiators have to do with how well-coordinated and composable its embedded algorithms are at scaling up various operational use case scenarios, he explained.

Claude Opus 4.7 is a “technically inclined” platform requiring a fair amount of deep personalization to fine-tune prompts and generate work outputs, he noted. It retains a strong lead over rival Google Gemini in terms of applied engineering use cases, even though Gemini 3.1 Pro has a larger context window (2M tokens versus Claude’s 1M tokens), although, he said, “certain [comparable] models do tend to converge on raw reasoning.”

The 4.7 update moves Opus beyond basic chatbot workflows, and positions it as more of “a copilot for complex, technical roles,” Levy noted. “It’s more capable than ever, and an even better copilot for knowledge workers.” At the same time, it poses less risk, making it a “carefully calculated compromise.”

He also pointed out that the Opus 4.7 release comes just two months after Opus 4.6 was introduced. That itself is “a signal of just how overheated the AI development cycle has become, and how brutally competitive the market now is.”

A guinea pig for Mythos?

Last week, Anthropic also announced Project Glasswing, which applies Mythos Preview to defensive security. The company is working with enterprises like AWS and Google, as well as with 30-plus cybersecurity organizations, on the initiative, and claims that Glasswing has already discovered “thousands” of high-severity vulnerabilities, including some in every major operating system and web browser.

Anthropic is intentionally keeping Claude Mythos Preview’s release limited, first testing new cyber safeguards on “less capable models.” This includes Opus 4.7, whose cyber capabilities are not as advanced as those in Mythos. In fact, during training, Anthropic experimented to “differentially reduce” these capabilities, the company acknowledged.

Opus 4.7 has safeguards that automatically detect and block requests that suggest “prohibited or high-risk” cybersecurity uses, Anthropic explained. Lessons learned will be applied to Mythos models.

This is “an admission of sorts that the new model is somewhat intentionally dumber than its higher-end stablemate,” Levy observed, “all in an attempt to reinforce its cyber risk detection and blocking bona fides.”

From a marketing perspective, this allows Anthropic to position Opus 4.7 as an ideal balance between capability and risk, he noted, but without all the “cybersecurity baggage” of the limited availability higher-end model.

Mythos may very well be the “ultimate sacrificial lamb” at the root of broader Opus 4.7 mass adoption, Levy said. Even in the “increasing likelihood” that Mythos is never publicly released, it will serve as “an ideal means of glorifying Opus as the one model that strikes the ideal compromise for most enterprise decision-makers.”

Palanichamy agreed, noting that Opus 4.7 could serve as a public-facing guinea pig to live-test and fine-tune the automated cybersecurity safeguards that will ultimately “become a mandatory precursory requirement for an eventual broader release of Mythos-class frontier models.”

The European Commission this week requested, but did not order Google to allow third party search engines in Europe access to its search data as a means to comply with the Digital Markets Act (DMA), legislation the Commission describes as a law designed to “make the markets in the digital sector fairer and more contestable.”

Google was sent a set of proposed measures on Wednesday that, according to a release, would grant third party search engines, including Qwant from France, Mojeek, based in the UK, swisscows from Switzerland, and Ecosia, Good, and metaGer, all headquartered in Germany, the ability to access search data, such as ranking, query, and click and view data “on fair, reasonable and non-discriminatory terms.”

In a statement, Teresa Ribera, executive vice-president for Clean, Just and Competitive Transition with the Commission, said that the decision “sets out the specifications we expect Google to follow to comply with its obligations under the [DMA]. Data is a key input for online search and for developing new services, including AI.”

The measures themselves cover several areas, including the scope of the search data Google must share, the means and frequency by which it must happen, and parameters for “setting fair, reasonable and non-discriminatory prices for search data.”

Move ‘far exceeds DMA’s original mandate’

In response to the Commission’s request, Clare Kelly, senior competition counsel for Google, said Thursday in a statement, “hundreds of millions of Europeans trust Google with their most sensitive searches, including private questions about their health, family, and finances, and the Commission’s proposal would force us to hand this data over to third parties, with dangerously ineffective privacy protections.”

The company, she said, “will continue to vigorously defend against this overreach, which far exceeds the DMA’s original mandate and jeopardizes people’s privacy and security.”

Phil Höfer, board member of SUMA-EV, which develops and runs MetaGer, said, “the planned measure might help with optimizing and developing European competitors to Google’s search service, but is not what’s needed most at this time. As long as the Commission isn’t planning on forcing Google to share their index data as well, this will not do much.”

Even better, he said, would be for the Commission “to decide to continue funding the European Open Web Index and allow European actors to build a competing infrastructure. We are convinced that without a European index, the EU will not be able to compete with American search engine giants.”

Forrester Senior Analyst Dario Maisto said the decision from the Commission is “not too timely but definitely in line with the measures Europe needs to free up businesses and citizens from risky dependencies on foreign organizations, vendors, and technologies. The final outcome is truly uncertain, though: one thing is to provide access to data to other players, one other thing is to modify users’ behaviors. We have to remember that the synonym for doing a search on the internet is actually: Google it.” 

Brian Jackson, principal research director at Info-Tech Research Group, said that opening Google’s search data to third parties could make search more specialized again, especially in high-value verticals where users want results tailored to a specific industry or service need.

Enterprise digital teams, he said, may need to optimize for multiple discovery environments rather than relying just on Google alone, and software buyers could see more choice as search and intelligence vendors build on shared data.

In addition, said Jackson, “it could revive domain-specific search models, but I think a more fragmented search ecosystem might raise manipulation risks, fraud, and poisoned results. That would make governance and monitoring much more important.”

Sanchit Vir Gogia, chief analyst at Greyhound Research, noted that, in terms of the impact on enterprises if Google shares search data under DMA, “this is being framed as a competition move, but that is not where the real impact sits. What is actually shifting here is control over how enterprise information is interpreted by machines.”

Definition of optimization is changing

For a long time, he said, “enterprises have quietly relied on the stability of a dominant discovery layer led by Google. That stability shaped everything from how content was written to how digital performance was measured. What is changing now is not just who has access to data, but how many systems can interpret that data.”

Gogia pointed out, “as alternative engines improve and start to matter, enterprises will find themselves operating in an environment where the same content can be surfaced differently, depending on which engine or AI system is doing the interpreting. That creates inconsistency, and over time, inconsistency becomes risk.”

There is, he said, also a deeper shift underneath all this: “Search is no longer just about helping users find information. It is increasingly the layer that feeds AI systems, copilots, and automated decisions. Once that layer fragments, enterprises no longer have a single reference point for how they are represented externally. That loss of coherence is subtle at first, but it builds into something much more material.”

Addressing the question of whether or not enterprises will need to optimize for multiple algorithms, he said, “the short answer is yes, but the bigger point is that the definition of optimization itself is changing. Enterprises are moving away from a world where they could tune for one dominant system into one where relevance is decided differently across multiple engines that do not follow the same rules.”

Search engines such as Qwant, Ecosia, and Mojeek, “each approach indexing and ranking differently,” Gogia said. “Some rely on their own infrastructure, others blend multiple data sources. The result is that the same piece of content can behave very differently across environments, even when nothing about the content itself has changed.”

What complicates this further, he said, “is the rise of AI-generated answers. Enterprises are no longer competing for links, they are competing to be included in summaries that may not even reveal where the information came from. That shifts the focus away from keywords and toward clarity, context, and credibility. The organizations that do well will be the ones whose content holds up across systems, not just within one.”

Interested parties have until May 1 to submit views on the proposed measures prior to a final decision, which will be binding on Google and must be adopted by July 27.