Kategorie

For years, Microsoft has hyped Windows 11 cas an OS with AI, and the company is finally putting the building blocks in place for that transformation.

Microsoft execs shared examples of how the company is integrating AI in Windows 11 at its Build event earlier this month, highlighting how AI models and agents will make the OS smarter, allowing users to interact with it using natural language and intent.

Specifically, Windows 11 PCs will provide unmetered intelligence so users can run AI for free without a network connection. “No token cost. No sensitive data leaves the device. It also reduces latency,” Anastasiya Tarnouskaya, product manager for Windows ML, said during a Build session.

Hardware makers introduced AI-capable hardware before the applications were available. But Tarnouskaya said more than 500 million PCs are already running local AI workloads. “Thanks to recent advancements in AI models, hardware, and the software stacks that run them, today, every Windows PC is becoming increasingly AI-capable,” she said. 

The AI experiences are blended into apps and the Windows UI, not working only as chatbots such as those offered by ChatGPT or Gemini. 

Microsoft Office, Photos and Teams already use on-device AI capabilities, with Outlook, for instance, summarizing emails using Microsoft’s Phi Silica model and a GPU on the PC.

“And it’s not just developers that are betting on local AI…, [companies] from Adobe to WhatsApp are building some incredible local AI-powered experiences.” Tarnouskaya said. Other early adopters include Canva, Affinity, and Speechify.

AI apps for Windows 11 proliferated after Microsoft shipped Windows ML last fall, she said. (Windows ML helps developers create offline AI applications without accessing cloud models. It maps applications, localized AI models and hardware such as GPUs and neural processors.)

Windows ML is part of Microsoft’s “Foundry” portfolio of products, which includes Foundry Local for running open-source models on Windows devices, and Windows AI APIs that automate tasks such as conversation summarization, speech recognition, and video upscaling.

Microsoft is also turning to AI agents to change how users interact with Windows 11. Users can describe a task through natural language, and a long-running agent will get to work and complete the action. “Windows is evolving into a platform where natural language can map to real system outcomes,” said Samantha Song, product manager for Windows at Microsoft.

Song demonstrated how users could just tell or type how they want to personalize colors, wallpaper, or menus, and the agent will do it. “There is no manual set up against themes, setting or lighting. The system treats it as one coherent action,” Song said.

For the effort to succeed, developers will need to create a skills file that maps how an agent behaves. That skill can then be reused over and over again, Song said.

“At the enterprise level, you could imagine a world where a user switches into a secure finance mode, and the system aligns apps, access boundaries and environment automatically,” Song said.

Microsoft also demonstrated how OpenClaw can be used to create personalized agents to run Windows functions.

At Build, LLMware.ai demonstrated an agent on a Qualcomm laptop that collects Jira issues in real-time, summarizes them locally, and emails daily summaries of top issues to the team. The agent runs automatically without prompting.

“You can get optimized performance on the NPU [neural processing unit] by running the model locally…and you also implement a scheduled run of your automated agents,” said Darren Oberst, co-founder of LLMWare.ai.

Samsung, Lenovo and others are rolling out — albeit slowly and carefully — agentic AI features under the moniker of ‘personal AI,’” said Leonard Lee, principal analyst at Next Curve. “The problem is ensuring safe deployment,” he said.

Microsoft’s efforts to embed AI in Windows will force enterprises to rethink hardware strategies, said Jack Gold, principal analyst at J. Gold Associates. And since AI chips excel at different tasks, Microsoft will have to support multiple chips to offer choice to enterprises, he said.

“We recommend — and others do, too — that any new PC purchases, especially for enterprise, be done with this in mind and purchase AI PCs during any upgrade cycle,” Gold said.

AI is here, and we must help workers adapt: That’s the response of a new non-profit organization to the ongoing debate about whether AI will destroy jobs and cause catastrophe or take the economy to new heights.

Launched Thursday, Raise Us is a nonpartisan national organization that says it will partner with state governors, employers, workers, and training organizations to help the US workforce make a successful transition to an AI economy. It says it will design and pilot new corporate incentives to retrain and redeploy workers, develop new approaches to support people through job transitions, and create new training models tied to changing employer demand.

It has already raised more than $500 million and hopes to bring that up to $1 billion in multi-year commitments from philanthropists and industry sources. Anchor partners include Amazon, Microsoft, Anthropic, and the OpenAI Foundation, and more than two dozen other organizations including Cisco, IBM, ADP, AMD, and Deloitte have also signed on.

The organization is led by two former US state governors, Gina Raimondo, who will serve as CEO and co-chair, and Eric Holcomb, co-chair, and its initial government partnerships are with the states of Arkansas, Connecticut, Maryland, and Utah, which, it said, “will serve as the first proving grounds for outcome-driven pilots.”

However, analysts are skeptical that the initiative is anything other than a public-relations effort for the AI industry.

Jason Andersen, VP and principal analyst at Moor Insights & Strategy, said the organization’s initial communications are more marketing and posturing than strategy and action. “It also does not help that there are many CEOs involved who have already displaced thousands of workers due to AI,” he said.

The organization has chosen to work with state governments rather than the US federal government but even partnering at the state level may be too broad, he said, given that the industry makeup varies from state to state, with some leaning towards knowledge workers, while others are more manufacturing-focused.

AI-washing

Technology analyst Carmi Levy said Raise Us’s mission is laudable, but, “Unfortunately, for all its slick website and optimistic-sounding copy, Raise Us is possibly the highest-profile example of AI-washing thus far, a well-intentioned but ultimately futile attempt to illustrate that something is being done to cut through the uncertainty and empower the workforce of tomorrow. In reality, it’s a convenient vehicle for a diverse range of stakeholders, including state governments, educators, financial institutions, and companies like Amazon, Anthropic, Microsoft, and the OpenAI Foundation, to virtue-signal that they’re doing their utmost to solve the AI employment crisis and mitigate the impact of this generational technology deployment on everyday workers.”

Like Andersen, he noted that since many of those same companies have already had multiple rounds of tech-driven layoffs, “It’s more than a little rich for them to be joining an initiative whose mission revolves around protecting workers from the worst risks associated with AI.”

But, he said, “If Raise Us can deliver on its lofty promises, it could be an important first step in the great AI workforce transition. Otherwise, it could end up being little more than a high-profile exercise in creating plans and reports and conference panels with little on-the-ground execution to show for it. I’ll believe in its mission when middle-aged, mid-level administrative workers are realistically redirected into new AI-enabled career paths without a massive pay cut.”

Andersen, too, said he’s keeping an open mind, “but something more measurable and specific is in order, or this just looks like a public relations effort.”

Have you seen “The Founder”?

It’s the story of McDonald’s and how Ray Kroc (played by Michael Keaton) transformed the company from a local burger joint to a global landlord. According to the movie, Kroc’s accountant gave him the revelation: “You’re not in the burger business. You’re in the real estate business.”

When brothers Richard and Maurice McDonald transformed their San Bernardino, CA barbecue restaurant into a fast-food burger joint in 1948, their model was to make and sell their own burgers. They would succeed or fail based on making better food products than other restaurants. 

By the time Kroc bought out the McDonald brothers in 1961, the new model was leasing real estate to other people, and those other people would make the food. Whether the original San Bernardino McDonald’s succeeded or failed became irrelevant to the success of the McDonald’s corporation. 

SpaceX has done the same thing. Its San Bernardino location, i.e. xAI, can now succeed or fail without affecting the success of the parent company, which is SpaceX. 

The company this week signed a compute lease with Reflection AI, a pre-revenue startup founded by former Google DeepMind researchers. Under the agreement, Reflection pays $150 million per month for use of the Nvidia GB300 chips housed at Colossus 2, SpaceX’s expansion facility in Memphis, TN. If the lease runs for the full term, SpaceX as a landlord stands to make around $6.3 billion. (Reflection AI has shipped open-weight models, but has no widely adopted frontier model yet, and was reportedly raising capital at a $25 billion valuation.) 

Earlier this month, an S-1 filing revealed that Google agreed to pay SpaceX approximately $920 million per month for 32 months. SpaceX stands to make around $30 billion. And last month, SpaceX disclosed that xAI made a big deal with Anthropic that could bring in to SpaceX as much as $45 billion in revenue.

Regardless of whether xAI — or, for that matter, Reflection, Google, or Anthropic — succeeds or fails, SpaceX still wins. 

Like McDonald’s (which developed its super efficient burger-building system in order to succeed with its San Bernardino location and ended up using that system to succeed as a landlord), SpaceX is using the Colossus infrastructure it built for xAI’s Grok to succeed as an AI landlord.

Grok might succeed, or it might fail. But SpaceX makes bank if Grok’s competitors pay their rent. 

That makes Elon Musk, who is the CEO of SpaceX, the Ray Kroc of AI. 

Nvidia is Mayor McCheese

Colossus originally went online in July 2024, powered by 100,000 Nvidia H100 Hopper GPUs housed in a Supermicro liquid-cooled HGX H100 chassis. The company doubled that to 200,000 GPUs within a short time and it now comprises more than 220,000 Nvidia GPUs including H100, H200, and next-generation Blackwell-class accelerators. The entire fabric runs on Nvidia’s Spectrum-X Ethernet platform, specifically the Spectrum SN5600 switch built on the Spectrum-4 ASIC. 

Also: Nvidia is also heavily invested in companies that are renting compute power on Colossus. The company invests in Anthropic and Reflection AI — and, for that matter, xAI itself and, therefore, SpaceX. 

Nvidia has positioned itself as the Mayor McCheese of the AI industry, collecting taxes at every node of the AI economy. It supplies the GPUs and networking fabric that every frontier lab must train on and collects hardware revenue from the winner’s rivals, even as it profits from the winner’s success. 

So whether Anthropic, Google, Reflection AI, xAI, or some yet-unformed lab produces the dominant model, the computing power was bought from Nvidia and the landlord’s machine was built from Nvidia silicon. 

And Apple is the Hamburglar

Apple’s AI strategy is even more brilliant than Nvidia’s. 

It’s built on a three-tier routing system. When you ask Siri to do something, a built-in orchestrator in the operating system decides how complex the task is. According to third-party estimates, around 85% of requests are handled on your Apple device by Apple’s own small, efficient models. (It does things like summarizing text, prioritizing notifications, cleaning up photos, or suggesting replies.) Roughly 12% of all queries get sent to Private Cloud Compute, Apple’s own server infrastructure running Apple’s larger models on Apple silicon in Apple-owned data centers. Only the hardest 3% of queries get routed to an external partner model.

This design lets Apple avoid the ruinous cost of training a frontier model from scratch. Microsoft, Google, Meta, and Amazon each spend tens of billions of dollars per year on GPU clusters, energy, and research teams to build and run trillion-parameter models. Apple doesn’t. Its own models are deliberately small and run on chips Apple already sells you, so the inference cost is basically absorbed into the device. It only needs a frontier model for that tiny sliver of hard queries, which is where the partnership strategy kicks in.

While frontier labs are collectively spending trillions to build AI infrastructure, Apple is paying Google a mere $1 billion per year to license a custom Gemini model that powers the rebuilt Siri and Apple Intelligence’s complex-query path.

The reason Apple can swap partners is the Foundation Models framework, a native Swift API with a published LanguageModel protocol that any provider can implement. Google’s Gemini conforms to it. Anthropic’s Claude probably conforms to it. Any future model can theoretically conform to it. Apple’s orchestrator routes to whatever model fits the interface, so switching providers means changing routing logic, not rebuilding the whole system. 

Apple profits through several channels. Apple Intelligence requires recent hardware, driving upgrade cycles. Advanced features push users toward higher iCloud storage tiers. 

And so while everyone else is investing trillions, creating what is essentially debt that has to be repaid somehow, Apple is mainly just collecting billions without the massive investments needed by the frontier model companies. 

The AI industry has been telling itself a story: that the companies building the best models will win, that intelligence is the product, that the chatbot with the most capabilities and the cleverest training run will capture the market. That story is wrong. Some of the companies building the best models are tenants. The companies that rent out the compute are landlords. 

Ray Kroc would recognize SpaceX’s strategy immediately. The burger doesn’t matter; the land does. Right now, the most valuable land in the world isn’t in Silicon Valley. It’s a data center complex in Memphis full of hundreds of thousands of GPUs. 

The man who owns it just realized that he’s not in the AI business at all. He’s in the real estate business. (And he’s probably lovin’ it.)

AI disclosures: I don’t use AI for writing. The words you see here are mine. I used a few AI tools via Kagi Assistant (disclosure: my son works at Kagi) as well as both Kagi Search and Google Search as one part of my fact-checking for this column. I used a word processing product called Lex, which has AI tools, and after writing the column, I used Lex’s grammar checking tools to hunt for typos and errors and suggest word changes. Why I disclose my AI use and encourage you to do the same. 

An active phishing campaign has been targeting hotel and other hospitality organizations across Europe and Asia since April 2026, using photo-themed ZIP files to drop a Node.js implant and dig into front-desk machines, Microsoft says. The company has not attributed the activity to a known threat actor, and the operators' end goal is still unclear. The lure plays to how hotels work. Swati Khandelwalhttp://www.blogger.com/profile/[email protected]

Russian authorities used Cellebrite's UFED forensic tools to break into the iPhone of detained opposition activist Andrey Pivovarov in June 2021, three months after Cellebrite said it would stop selling its tools and services to Russia and Belarus. The finding, published June 25 by the Citizen Lab, rests on two things that rarely line up: traces on the phone itself and an official Russian Swati Khandelwalhttp://www.blogger.com/profile/[email protected]

**Nejlepší hračky a stavebnice za vysvědčení ** Vybíráme herní počítač na prázdniny ** Nejlepší méně známé aplikace pro cestovatele.

** Miliardy telefonů s Androidem tvoří globální síť pro detekci zemětřesení ** Systém poskytuje včasné varování před otřesy miliardám lidí po světě ** Tato technologie se stále vyvíjí a má obrovský potenciál pro výzkum

The Russian state-sponsored threat actor known as Turla has been attributed to a previously undocumented .NET backdoor called STOCKSTAY that has been deployed against government and military organizations in Ukraine, and entities that have an interest in Italian foreign policy. Describing the Windows backdoor as continually developed by the hacking group, Google Threat Intelligence Group (Ravie Lakshmananhttp://www.blogger.com/profile/[email protected]

Anthropic appears to be testing Claude Cowork support on mobile, allowing you to manage long-running Claude tasks from your phone. [...]

Authorities in Poland have arrested four members of an organized cybercrime group accused of breaching telecommunications partners and hijacking email accounts to carry out SIM-swapping attacks. [...]

Threat actors are increasingly abusing Shop, the order-tracking app from Shopify, by adding fake purchase receipts in users' order histories to trick them into providing sensitive data or installing remote access software. [...]

In February 2024, Notion bought Skiff, an encrypted email and productivity software startup. Within a year, Notion shut down Skiff’s email service (taking @skiff.com email addresses with it). And in April 2025, the San Francisco-based company released Notion Mail, a Gmail client primarily built by people who joined Notion through the Skiff acquisition. Today, Notion announced that it’s shutting down Notion Mail, effectively killing what little remained of Skiff email.

In an X post (first spotted by 9to5Mac) today, Notion said that it will shutter the Notion Mail “inbox across web, desktop, and iOS on September 22.”

The post claimed that most Notion users don’t use email clients anyway and instead rely on AI agents to handle their electronic correspondence. It reads:

Read full article

Comments

Microsoft has quietly extended its free Windows 10 Extended Security Updates (ESU) program for consumers by an additional year, allowing enrolled devices to continue receiving security updates until October 12, 2027. [...]

Microsoft’s controversial claim that its Majorana chip program will make possible a scalable quantum computer by 2029 has been thrown into new doubt by a scientific paper that questions whether the company has correctly interpreted its own experimental evidence.

According to a peer-reviewed paper by Dr. Henry Legg from the University of St Andrews, published this week in Nature, Microsoft’s Topological Gap Protocol (TGP) framework, designed to infer the existence of quantum states in theorized Majorana particles, is flawed.

“Last year Microsoft claimed they had built the equivalent of a precision Swiss watch. However, when I opened the case to examine the mechanism, I found what looked like a chaotic jumble of mismatched parts,” said Legg.

He believed the results gathered from Microsoft’s TGP software data analysis could also be explained by other effects, as well as being skewed by the data chosen for analysis. Because of this, he believed the company’s researchers had jumped to the wrong conclusions.

“Something was making noise, but it didn’t look like the breakthrough Microsoft had claimed. Despite the headlines, the vast majority of scientists in the field were skeptical of Microsoft’s claim from the start; my critique simply backs up that skepticism in the scientific record,” he said.

Topological qubits

The ability to create Majorana ‘zero modes’ that resist the errors suffered by traditional qubit-based designs is fundamental to Microsoft’s entire quantum computing strategy, stretching back two decades. This, of course, assumes the existence of subatomic Majorana fermions, named after the Italian physicist who first proposed them in 1937. To this day, they remain only theoretical.

In 2018, Microsoft said its researchers had detected evidence of their existence, an apparently major breakthrough it was forced to retract when the data was successfully challenged. Nature’s editors subsequently backed this up with the blunt note: “The results in this manuscript do not represent evidence for the presence of Majorana zero modes in the reported devices.”

Despite the setback, Microsoft persisted, in 2025 publishing a new paper in Nature that claimed the company had worked out how to exploit the principle in a new “transistor for the quantum age,” the Majorana 1 chip.

Earlier this month, the company launched its successor, Majorana 2, claiming that AI had helped to achieve a 1,000-fold improvement in reliability compared to the earlier chip. “With this progress, the team now expects to achieve a scalable quantum computer by 2029, cutting its original timeline in half,” the company announced.  

However, Microsoft again finds itself on the defensive after the latest criticism from Legg, who said, “I am simply reflecting what most in the field felt from the initial announcement. I felt that I needed to put these concerns into a formal scientific critique. It is good that it has now been peer-reviewed and published.”

His criticism relates to the transport data system, not the raw experimental data itself, which Microsoft had yet to make fully public in a way that would allow independent analysis.

Microsoft still confident

Microsoft said by email that it remains confident that the Majorana has brought usable quantum computing a step closer, pointing to its collaboration with DARPA as part of the Underexplored Systems for Utility-Scale Quantum Computing (US2QC) program.

“We stand by our results and our roadmap,” said technical fellow and vice president of quantum hardware, Dr. Chetan Nayak. “At the end of the day, success is the delivery of a scalable quantum computer. We are confident in our ability to execute against our roadmap.”

“Skepticism and rigor are hallmarks of the scientific process, which we appreciate and have supported from various academics,” he added. “We have participated in dialogue and our thorough rebuttal was accepted and published by Nature.”

Microsoft is not the only company researching quantum computing hardware, with Google, IBM, and Amazon also working on designs. But even if the hardware matures on the schedule claimed by Microsoft,  many believe the gap between technology and implementation inside enterprises could be more of a slow burn than a sudden jump forward.

This article originally appeared on NetworkWorld.

Apple’s threatened product pricing hike arrived Thursday with steep increases landing across all its products (except the iPhone) — and rapidly accelerating RAM prices are to blame.

Apple CEO Tim Cook had warned last week about the need for price increases, with Bloomberg reporter Mark Gurman subsequently telling us these were “imminent.”

The increases have been imposed as a result of the massive demand for memory and storage components generated by the rapid investment in AI servers. Prices have increased to meet this demand, with vendors shifting manufacturing to the advanced memory modules demanded by AI companies. In doing so, they have not invested in additional production capacity, further exacerbating the demand/supply imbalance.

“Unfortunately, price increases are unavoidable,” Cook told The Wall Street Journal. “We’re doing our best to mitigate the huge increases that are being passed to us, and we’ve been trying to shield our customers from the increases, but the situation has become unsustainable.”

The price increases are going to hurt

The price hikes even affect products Apple hasn’t updated in years, such as the HomePod and Apple TV. The extent of the increase is big in some cases — the Apple TV is up 55% while an M3 Ultra Mac Studio sees a 32.5% rise. While I had hoped any higher prices would be aimed at Apple’s high-end products, the extent and scale of the changes suggests the enormity of the component price increases the company has been struggling with.  

“We haven’t seen anything like this in modern Apple,” said Neil Cybart at Above Avalon.

Here’s what starting prices look like now:

Macs

• MacBook Pro, up $300 to $1,999.
• MacBook Air, up $200 to $1,299.
• MacBook Neo, up $100 to $699.
• iMac, up $200 to $1,499.
• Mac mini (M4 Pro), up $200 to $1,599. 
• Mac Studio (M4 Max), up $500 to $2,499.
• Mac Studio (M3 Ultra), up $1,300, to $5,299.

iPads

• iPad Air, up $200 to $749.
• iPad Pro, up $200 to $1,199.
• iPad, now starts at $449.
• iPad mini, now $599.

Other hardware

• Vision Pro, up $200 to $3,699
• HomePod mini, up $30 to $129.
• Apple TV, up $70 to $199.

There were no immediate changes to the iPhone, Apple Watch and AirPods — but no guarantees, either, that their prices won’t also jump.

This is AI-flation

The scale of these increases is like a dose of cold water on the already tepid optimism of investors. This was quickly reflected by a big sell off on Apple stock, which was down more than 6.5% at one point today. And the impact was contagious across the sector as investors realized that if Apple was forced to increase prices even at the lower end of the market it so recently secured, it’s likely other manufacturers will be forced to do the same. This can only further dampen market optimism.

Concerning the MacBook Neo, Francisco Jeronimo, IDC vice president for client devices, said: “The $100 increase takes the entry model from $599 to $699, close to a 17% increase, one of the sharpest percentage hikes in the announcement, and it hits one of Apple’s best-selling, most price-sensitive laptops right as it is winning share. Raising the price of its fastest-selling entry product indicates that Apple believes demand will stay strong, given the value the product offers. It also signals Apple is willing to give up some unit growth at the bottom of the range to defend margin, and it widens the gap the Neo had just closed against Windows entry-level machines.”

The fact of the matter is that consumers are effectively paying an inflationary fee for the benefit of the ongoing global AI rollout. That rollout is already consuming vast quantities of investor cash, despite the shaky business plan of existing incumbents. 

“We have never seen a component price increase this much, this quickly,” Apple ​said in a statement. “We have shielded our customers from these increases so far, but we have now reached a point where we need to begin ⁠raising prices on a number of products, including today’s increases for iPad and Mac.”

Apple and the entire industry will not want to maintain prices at these levels if they can avoid it. That’s going to mean the entire tech industry will now begin leaning on memory manufacturers such as Micron to do more about building out production capacity, or at least promising to do so. Of course, the other hope might be that investors finally see the folly of the AI business model and slow down deployment, though that seems unlikely.

“We know this is not welcome news, and we are working tirelessly to find solutions,” the company said.

Please join me on social media at BlueSky,  LinkedIn, or Mastodon, even better, please subscribe to The Core for your daily collection of human-curated Apple News.

A newly discovered macOS malware dubbed "Gaslight" is designed to confuse AI-assisted malware analysis tools by hiding prompt injection strings and fake debugging data within the executable. [...]

A major sports piracy ring linked to the illegal PirloTV streaming platform has been disrupted in an action that targeted 44 domains. [...]

The Bluekit phishing-as-a-service platform continues to evolve with nearly 70 new hostnames identified over the past week and by adding browser-in-the-middle capabilities for improved data theft. [...]

**Přihlašování pomocí uživatelského jména na Spotify zruší. **Od 1. září 2026 se budete moct přihlašovat jen e-mailem. **K dispozici zůstane přihlášení přes služby třetích stran.

An analysis of a popular Google Chrome ad block extension for YouTube has uncovered the ability to execute arbitrary JavaScript code. According to Island, the extension, named Adblock for YouTube (ID: cmedhionkhpnakcndndgjdbohmhepckk), has more than 10 million installs and carries a Featured badge on the Chrome Web Store. The extension description states that it allows users to prevent web Ravie Lakshmananhttp://www.blogger.com/profile/[email protected]