Agregátor RSS

The China-nexus cyber espionage group tracked as Armory Castle (previously UNC5221 or Warp Panda) exploited a newly disclosed security flaw in BeyondTrust Remote Support (formerly Bomgar) as a zero-day in late August 2026. The vulnerability, tracked as CVE-2026-102764 (CVSS score: 9.3), is a case of improper origin validation in an internal messaging component that could be exploited by a Ravie Lakshmananhttp://www.blogger.com/profile/[email protected]

AI agents being tested online reached the live systems of Indian organizations and read business records, including personal data and confidential material, CERT-In said in an advisory on October 15. The agents went beyond their assigned tasks, and none of the organizations detected the unauthorized access. After internal reviews, the AI companies whose models were being tested

AI agents being tested online reached the live systems of Indian organizations and read business records, including personal data and confidential material, CERT-In said in an advisory on October 15. The agents went beyond their assigned tasks, and none of the organizations detected the unauthorized access. After internal reviews, the AI companies whose models were being tested Swati Khandelwalhttp://www.blogger.com/profile/[email protected]

Nahlédneme do budoucnosti a podíváme na jednu z nejvýkonnějších integrovaných grafik u procesoru. Dorovná dedikované mobilní i desktopové grafiky, ale vyplatí se?

BeyondTrust warned customers on Thursday to patch a critical flaw in its Remote Support (RS) that has been exploited in zero-day attacks since late August. [...]

V květnu jsme vás upozornili, že zavedení paušálního cla €3 za každý typ zboží v zásilce nebude konečná částka, protože se chystá ještě tzv. manipulační poplatek. Bude méně příjemný, než se čekalo…

Na všech herních platformách je každou chvíli nějaká slevová akce. Každý týden proto vybíráme ty nejatraktivnější, které by vám neměly uniknout. Pokud chcete získat hry zdarma nebo s výhodnou slevou, podívejte se na aktuální přehled akcí!

A suspected advanced threat actor has been observed exploiting a security flaw in Fortinet FortiOS to deliver a previously unreported memory-resident ARM64 malware known as PEERSTITCH. The attack involves the exploitation of CVE-2025-61624 (CVSS score: 5.4), a path traversal flaw in the command line interpreter of FortiOS, FortiPAM, FortiProxy, and FortiSwitchManager that could allow a Ravie Lakshmananhttp://www.blogger.com/profile/[email protected]

A suspected advanced threat actor has been observed exploiting a security flaw in Fortinet FortiOS to deliver a previously unreported memory-resident ARM64 malware known as PEERSTITCH. The attack involves the exploitation of CVE-2025-61624 (CVSS score: 5.4), a path traversal flaw in the command line interpreter of FortiOS, FortiPAM, FortiProxy, and FortiSwitchManager that could allow a

Tesla na unikátní prezentaci ukázala elektrické sportovní kupé Roadster. • Verze Thrust s tryskami na stlačený vzduch zrychlí na stovku pod sekundu. • Už jsou spuštěné rezervace, létající model Rocket bude jen na pozvánky.

Společnost Fairphone stojící za stejnojmennými mobilními telefony představila (pdf) bezdrátová špuntová sluchátka Fairbuds 2. Opravitelná, s vyměnitelnými bateriemi a tříletou zárukou. Cena 119 eur.

O víkendu 17. a 18. října lze na brněnském výstavišti navštívit s jednou vstupenkou dvě akce: Maker Faire Brno, festival inovací, kreativity a kutilství, a GameDev Connect, akci určenou pro všechny současné a hlavně budoucí herní vývojáře.

That ad you just scrolled past on your phone but barely spared a look or click? It could be tracking your movements and collecting your data, without you having to do a thing.

This “zero-click mercenary spyware,” dubbed “AdFection,” is increasingly exploiting a sprawling online advertising ecosystem ripe for abuse, according to researchers at The Citizen Lab. One top actor in the space, Discover-Tech, runs the Controlium platform that allows cyber criminals to “precisely profile and de-anonymize” millions of users, fingerprint them, and automatically infect their devices when an ad is displayed.

Troublingly, Discover-Tech is successfully masquerading as a legitimate ad company across the online ad sales environment, the researchers said, listed as a “certified” and “approved” vendor by the likes of Google and IAB Europe.

Over a four-month period, the company’s ads were shown on nearly 200,000 unique websites and in Android/iOS apps in dozens of countries. Furthermore, its ads have been presented through major networks like Google, Yandex, and Beeswax, and it even exploited the European General Data Protection Regulation (GDPR) consent frameworks to target victims in the EU.

The Citizen Lab researchers disclosed their findings to Google, IAB Europe, Beeswax, Vungle (Liftoff), and Eskimi, all of whom took action to disconnect Discover-Tech’s access to the digital advertising ecosystem.

“Discover-Tech’s mass abuse, privacy invasion, and facilitation for mercenary spyware attacks illustrates structural flaws with the online advertising system, which failed to detect or prevent the abuse,” the researchers warned.

How AdFection exploits a largely unprotected system

The online ad market operates largely via real-time bidding (RTB), auctions hosted by ad exchanges. Run by companies like Google, these take place “hundreds of billions” times a day, the researchers explained. When an ad box pops up on an app or website on a user’s device, it immediately triggers a bid request; advertisers then submit their offers, and the exchange grants a winner the ability to push their ad to the user within split seconds. These ads can contain images, video, and, most importantly to cybercriminals, code.

Demand-side platforms (DSPs) and supply-side platforms (SSPs) act as intermediaries in the process, and data about specific users and devices is shared across the exchanges so advertisers can decide whether they want to show specific ads to a specific individual, based on device identifiers and characteristics such as a user’s precise location, cookie ID, mobile ad ID, or IP address.

“The auction system is designed to enable targeting single individuals,” the researchers explained. “The totality of information gleaned from ad auctions is quite valuable.” Companies with access to auction data can track users without their knowledge, across cellular, Wi-Fi, and virtual private networks (VPNs).

Discover-Tech has effectively infiltrated and connected the spyware industry to this online ecosystem, which includes infamous Israeli mercenary spyware vendor Candiru. Its partners can deliver ads to “mass-profile unwitting phone users,” and silently infect the devices of select targets, according to the researchers.

When Discover-Tech wins a bid, one of its ads is pushed to a website or app that is open on a user’s device. Sometimes this is seen by the user; other times is merely renders in the background.

Many of these ads have been shown to devices on national security-sensitive networks, in security organizations including NATO, networks associated with critical infrastructure, and media organizations around the world. Using AdFection, Candiru’s spyware infected devices at international news network Al Jazeera.

The researchers emphasized that a Discover-Tech ad impression does not mean that a device has been infected with spyware. However, the fact that the company has been able to bid on and show ads across sprawling ad networks highlights a “systemic risk” warranting “urgent investigation.”

Lack of vetting

Discover-Tech’s ads carried no exploit at all, noted Frank Dickson of Dickson Research. “The ad was clean; that is the frightening part. Nobody clicked anything.”

Instead, each ad called home with the device’s advertising ID and IP address. In almost every case, the server sent back nothing, Dickson emphasized; only chosen targets received malicious code.

“Third-party verification scanners saw a harmless banner, because for nearly every viewer, it was one,” he said. This is exactly how tags slipped past verification.

Also alarming is the reach: Discover-Tech’s ads displayed in 75,373 apps and 120,768 websites over the course of four months, and had impressions on networks belonging to NATO, Intel, Microsoft, and Apple. While the report emphasizes that an impression is not an infection, the company’s targeting clearly reached enterprise networks, Dickson said.

Just as troublingly is the badge: Discover-Tech was listed as a certified or approved vendor for years. The report said that vetting often amounted to self-attestation, a box checked on a form. “A certification is only as good as the vetting behind it,” Dickson said.

Steps enterprises should take now

On managed browsers, ad blockers are necessary, but they’re not wholly sufficient, Dickson said. Many Discover-Tech bids were for ads inside mobile apps, typically overlooked by browser extensions. But clearly, desktops are at play too; Candiru targeted Al Jazeera’s Windows software.

Dickson advised enterprises to:

• Push a vetted ad-blocking extension to managed browsers through policy rather than leaving it to users to install it.
• Favor browsers with built-in tracking protection.
• Block ad and tracking domains at the DNS or secure web gateway layer; this reaches in-app ads on managed phones.
• Disable advertising identifiers on corporate mobile devices through mobile device management (MDM) and restrict ad-supported consumer apps on devices holding sensitive data.
• Turn on Apple’s Lockdown Mode for high-risk executives, journalists, and any users traveling to sensitive locations.
• Keep browsers and operating systems patched “The ad is only the delivery truck; the exploit still needs a flaw,” he pointed out.
• Watch traffic tied to ad impressions for odd follow-on requests.
• Finally, search domain name system (DNS) and proxy logs for the domains published in the report.

The advertising ID in particular deserves a hard look, Dickson pointed out. The Citizen Lab report says that blocking advertising identifiers alone would have stopped this attack method.

Troublingly, he noted, many enterprises spend fortunes managing their users’ identities, then let corporate phones hand that persistent identifier to anonymous auctions that run hundreds of billions of times a day.

“Every ad is someone else’s code running on your device,” Dickson said. “Treat the ad network like any other untrusted vendor.”

This article originally appeared on CSOonline.

NGINX has patched flaws that could let request data bypass a size limit or allow an encrypted session to be forged.

Although the Jev model rolled out last month by TypeSafe claims to have significant advantages over traditional LLMs, a security report published Wednesday argued that its streamlined approach makes it more vulnerable to some kinds of attack, compared to traditional LLMs.  

“In some cases, [Jev] is even more vulnerable to attacks, with both adversarial text attacks and prompt injection being able to disrupt the model’s outputs, resulting in incorrect and potentially devastating actions,” said the blog post from researchers at HiddenLayer. 

TypeSafe did not respond to multiple requests for comment. 

LLM, or something else?

There seems to be debate about whether Jev is technically a large language model (LLM), with TypeSafe suggesting it is not, but HiddenLayer saying that it appears to be one. TypeSafe has even coined a name for the category: System One, which TypeSafe described as “a new class of frontier models built to make fast, structured decisions that software can use directly.” Jev is the first—and thus far, only—named product categorized as System One.

Jev is a streamlined model designed to function machine-to-machine. Its approach allows it to perform tasks faster and to use fewer tokens. But Kasimir Schulz, the director of security research at HiddenLayer, the blog’s co-author, said that the accelerated approach robs Jev of some of the checks and balances that exist in traditional LLMs. This means that Jev can fall for some attacker tricks that current LLMs typically detect and refuse, he said in an interview. 

Despite this, “Jev is great. You should use it. It has a lot faster response time,” he said, with the caveat “it doesn’t have all of the protections” seen in traditional LLMs, so some attacks on it succeed that should have failed. 

“TokenBreak, for example, normally does not work with larger LLMs, but Jev is vulnerable to it out of the box,” Schulz said. 

Although simply inserting characters in front of important words, as in the original TokenBreak attack, did not yield much, HiddenLayer’s blog noted, “we did notice that increasing the number of perturbations and improving their targeting led to much higher attack success rates.”

But although traditional LLMs are typically able to see through such attacks, it pointed out, “System One models appear to be vulnerable. As a result, this class of decoder should be treated with caution, given that they are susceptible to both autoregressive LLM attacks and text classifier attacks, providing potential adversaries with greater optionality.”

The post cited some examples: 

“On a simulated checkout page, an attacker-controlled advertisement caused the browser agent to click away from the user’s purchase [and respond to the ad] despite instructions to treat page text as untrusted,” it said. “In a separate classification test, targeted TokenBreak-style changes lowered toxicity scores for prompts both models had initially flagged.”

The post argued that multiple risks are therefore associated with Jev. 

“A browser agent may immediately click a selected element. A moderation system may allow an input whose toxicity score falls below a threshold. A routing system may send a request to a different tool or workflow. Each of these turns a model judgment into an operational decision, and therefore, risk,” the post said. “When evaluating System One models, test those decisions in the application where they will be used. Include untrusted webpages, misleading task-related text, and modified inputs that preserve their meaning while changing their form. Monitor the selected actions and scores and put independent checks around outcomes that matter. Keep in mind that, under the hood, these are still language models at heart.”

Analysts agreed with both the report’s methodology and its technical conclusions, while also pointing out that a security vendor concluding that a new approach is not secure is somewhat self-serving.

Philip Harris, an IDC research director, observed that the methodology described in the blog post “is sound, and consistent with established AI red-teaming practice: building a controlled test environment and probing for adversarial manipulation and injection is exactly the right approach.”

But Gartner VP Analyst Nader Henein added that Jev, which was introduced on Sept. 15, is so new that he thinks some of the flaw reports were premature.

“Jev will go through the same maturity process as do all other models, including susceptibility to seemingly rudimentary manipulation attempts,” Henein said. “This model was only released last month, so being surprised that it has issues is similar to being surprised when your baby drools.”

However, Frank Dickson, principal analyst at Dickson Research, noted that the same attributes that make Jev so attractive also expose it to riskier behavior.

“Jev can only choose from options the developer puts on the menu,” he said. “It limits attack paths. However, if ‘click any element on the page’ is on the menu, the attacker’s ad is on the menu. That is an architecture decision, and the buyer makes it.”

He pointed out, “Jev makes AI decisions faster and cheaper, which also makes the wrong decisions faster and cheaper. Speed is a feature. It is not a security control.”

Independent technology consultant Steven Eric Fisher agreed.

“Constraining the format of a decision does not necessarily constrain the correctness, trustworthiness, or consequences of that decision,” he said, noting, “the simulated checkout scenario [in the post] illustrates how an agent can be redirected from an authorized task by information introduced through a less-trusted component of its operating environment. The fact that the agent was explicitly instructed not to trust page content, yet still responded to the malicious advertisement, is particularly relevant.”

This article originally appeared on InfoWorld.

An X.509 certificate names an identity and its public key; libraries can disagree over whether the issuer was allowed to certify that name.

A curl vulnerability can let a malicious upload server run commands on the computer sending it data.

Se začátkem příštího roku přijde několik daňových novinek, vesměs zaměstnance potěší. Podívejte se na jejich přehled.

Gavan Tredoux v knihe Recursing Ada ostro spochybňuje moderný obraz Ady Lovelaceovej ako prvej programátorky. Časť sporu však nie je kultúrna, ale technická: keď slávnu Note G nečítame ako artefakt, ale ako program.