Agregátor RSS

Removing a Linux trace instance should end its lifetime. An open tracefs reader can currently keep using that instance after another task removes it, creating a kernel use-after-free path in the tracing subsystem.

PaperCut has released a second emergency security update for two actively exploited vulnerabilities in its PaperCut NG and MF print management software after researchers discovered multiple ways to bypass the initial fixes. [...]

A Linux service may need broad system-call access while it starts, then only a smaller set while it handles requests. A single policy loaded before startup often has to keep every required call available for the service's entire lifetime.

A maximum-severity vulnerability in the GiveWP plugin for WordPress allows an unauthenticated attacker to execute arbitrary commands on the hosting server. [...]

Anthropic on Thursday launched the Model Hardware Standard, a new framework designed to enable the control of hardware using AI agents. With the new framework, AI agents will, for example, be able to operate robots or microscopess.

The Model Hardware Standard could also be useful for developing new drugs or calibrating the laser in a quantum computer, according to Reuters.

The company’s plan is to eventually release the new framework as open source, after it undergoes thorough testing.

Evropská komise uložila Googlu již čtyři pokuty ve výši přes miliardu eur. Ta první z roku 2017 činila 2,42 miliardy eur a týkala se zneužívání dominantního postavení ve vyhledávačích. Google upřednostňoval vlastní služby a srovnávače služeb před konkurencí, které odsouval ve výsledcích na méně ...

Malicious actors are exploiting a newly patched security flaw in PaperCut NG and MF to execute arbitrary code on susceptible instances, as the company released a fresh emergency fix with additional hardening. "This vulnerability gives an unauthenticated attacker remote control over PaperCut's trusted configuration, which could be used to execute arbitrary Java code inside the application's

Malicious actors are exploiting a newly patched security flaw in PaperCut NG and MF to execute arbitrary code on susceptible instances, as the company released a fresh emergency fix with additional hardening. "This vulnerability gives an unauthenticated attacker remote control over PaperCut's trusted configuration, which could be used to execute arbitrary Java code inside the application's Ravie Lakshmananhttp://www.blogger.com/profile/[email protected]

Recursive self-improvement, where AI continuously builds better versions of itself, might be harder than some hope.

There’s growing excitement in the AI industry about the idea that today’s leading models could build the next generation of the technology. But a new study recently found top AI agents struggle on the kind of genuinely open-ended research problems required to push the field forward.

Large language models have made rapid progress in many of the day-to-day jobs involved in machine learning research, such as writing code, generating and curating data, and running experiments. Last year, startup Sakana AI’s AI Scientist-v2 even managed to write a paper that cleared peer review for the prestigious International Conference on Learning Representations.

These advances have led to speculation that models are close to being able to build better versions of themselves with little human oversight—a process called recursive self-improvement. The idea underpins predictions that we may be on the verge of an intelligence explosion that could quickly lead to AI superintelligence.

In a recent paper, researchers put the idea to the test using a new approach they call shadow evaluations. This involves taking the research question from a high-quality, unpublished machine learning paper and asking AI agents to solve the problem. The original paper’s authors then grade the results. When the team tested Claude Opus 4.8 on two papers submitted to the prestigious machine-learning conference NeurIPS 2026, the authors rejected both.

“The papers were nowhere close to the mark when it came to being at the quality of a top AI conference,” Sayash Kapoor from Princton University, who co-led the study, told MIT Technology Review.

Previous efforts to get AI agents to do machine learning research have often targeted problems focused on engineering, such as reproducing previous research or training smaller models against a benchmark.

In the new experiments, the researchers challenged models with more open-ended tasks that required them to devise hypotheses, decide what evidence is needed to validate them, judge when a research direction was fruitless, and go back to the drawing board.

One research question was whether the personality traits a language model displays can be measured and adjusted by observing and editing its weights; the other attempted to detect when a model that works with tabular data has quietly stopped being reliable.

In each case, the AI researchers were given $3,000 of API credits, a budget for time on GPUs to run machine learning experiments, a dedicated Linux virtual machine, and unrestricted internet access. They were then given six days to produce a paper that could pass NeurIPS’ stringent peer-review criteria.

In both cases, the models got a good start. The agents surveyed the literature effectively, came up with opening hypotheses that mirrored those of the authors, and successfully ran hundreds of experiments.

But they quickly went off the rails. Although they could monitor their own use of time and their API and GPU budgets, they rushed through the process. One left 110 hours of unused time on the clock, and both failed to spend even 50 percent of their API budget.

Both agents also settled on a research direction within just 10 hours and failed to change approaches despite repeated negative feedback from another AI designed to review drafts of their papers. The reviewer identified problems the human authors would also flag in the final paper, but the models simply added caveats to their findings and ploughed on. Ultimately the papers received a “strong reject” and a “reject” decision from the human reviewers based on NeurIPS grading protocol.

The authors admit their approach has limitations. The reviewers knew AI had written the submissions, and some of the team are on record as doubting an imminent intelligence explosion. The original human-authored papers also took far longer than six days to produce and used many more GPU hours to reach their conclusions (though, as the researchers note, the models did not use their allocated budget in any case).

Nonetheless, the results suggest that today’s models still have some way to go before they can tackle the most challenging problems in machine learning research. Until that happens, the dream of recursive self-improvement is likely to remain a distant prospect.

The post Are We on the Verge of an Intelligence Explosion? Maybe Not. appeared first on SingularityHub.

Bez mobilu si už cestování nepředstavíme, jenže na každý krok potřebujete jinou aplikaci. Prozradíme vám, jaké nástroje zvolit na nákup jízdenek, proč se Google Maps a Mapy.com doplňují, nebo jak si poradit s cizími jazyky. Bez šťávy to nejde, takže probereme i nabíjecí strategii – od výběru ...

A 68-year-old has been sentenced in the U.K. to more than six years in prison for operating an illegal IPTV (Internet Protocol Television) service that generated £980,812 ($1.3 million) over three years. [...]

Google on Thursday announced new network security protections in Android 17 to bolster connection privacy, address cellular vulnerabilities, and safeguard the privacy of users' home networks. Topping the list is support for Encrypted Client Hello (ECH), a privacy standard that prevents networks from eavesdropping on which websites a user is visiting. "This new privacy standard works in tandem

Google on Thursday announced new network security protections in Android 17 to bolster connection privacy, address cellular vulnerabilities, and safeguard the privacy of users' home networks. Topping the list is support for Encrypted Client Hello (ECH), a privacy standard that prevents networks from eavesdropping on which websites a user is visiting. "This new privacy standard works in tandem Ravie Lakshmananhttp://www.blogger.com/profile/[email protected]

The U.S. Cybersecurity and Infrastructure Security Agency (CISA) on Thursday added a critical security flaw impacting ownCloud to its Known Exploited Vulnerabilities (KEV) catalog following reports that a Chinese-speaking threat actor weaponized the vulnerability to target a nuclear research body in the Philippines. The vulnerability, tracked as CVE-2023-49105 (CVSS score: 9.8), is a case of

The U.S. Cybersecurity and Infrastructure Security Agency (CISA) on Thursday added a critical security flaw impacting ownCloud to its Known Exploited Vulnerabilities (KEV) catalog following reports that a Chinese-speaking threat actor weaponized the vulnerability to target a nuclear research body in the Philippines. The vulnerability, tracked as CVE-2023-49105 (CVSS score: 9.8), is a case of Ravie Lakshmananhttp://www.blogger.com/profile/[email protected]

An important but potentially fragile trend is emerging in the data being reported by smartphone industry trackers: while painful component price increases are denting overall phone sales, the iPhone’s market share continues to increase. 

This likely reflects two key truths:

• Apple has fought to keep iPhone prices as stable as possible; though it has raised the price for its other products, it has not yet done so for its all-important smartphone.
• The company has built a great reputation for making smartphones that keep going for years and hold value better than rival platforms, though the looming price increase could put that perception to the test.

As prices rise, quality matters more

To most people, iPhones offer a perceived value. So, when the chips are down, consumers still gravitate toward them. They recognize the quality of these devices and, as the cost of other smartphones continues to increase, see iPhones as a more valuable alternative. 

Not only that, of course, but with the latest ‘e’ series iPhones, Apple is punching its way into the mid-range/value smartphone market, even as incumbents are forced to raise prices because of the AI-driven memory cost inflation. 

Smaller vendors are also being forced to raise prices or quit markets, with many reducing the number of devices they offer or leaving specific markets entirely. They are far more affected by memory price inflation than Apple, because they don’t order components at the same scale. 

“The 2026 decline reflects more than temporarily weak demand,” said Counterpoint analyst Yang Wang. “Higher component costs are pushing manufacturers to remove products and configurations that are no longer economically viable, particularly at lower price points.”

This is a global trend

IDC tells us to expect a record 16.7% decline in smartphone shipments in 2026 because of the memory crisis, which is pushing smartphone ASPs up 27.6% this year to around $581. (The researchers also expect Apple to inject some growth into this market when it introduces its folding iPhone). They point out also that while total shipments are falling, the value of the market is increasing — because prices are climbing. And they note that while low-end Android devices are disappearing, the premium end of the smartphone market (itself dominated by Apple and Samsung) remains more resilient.

“The era of the cheap smartphone has ended,” said Francisco Jeronimo, IDC vice president for worldwide client devices. “From here, the winners will be the vendors with the scale and supply leverage to hold demand at prices consumers have never had to pay before.”

The picture is much starker on a platform basis, as Android devices account for almost the entire market decline; its share fell seven percentage points in a single year. Meanwhile, iOS share grew almost four percentage points to a record-high 23.6%.

The pattern is being repeated globally — nation by nation and no matter which analyst you choose to track. Counterpoint expects Apple to gain relative share, even as the market declines. That’s true in most regions; the same analyst reports that while the MEA smartphone market fell 10%, the iPhone gained 28% share. It’s the same in Europe, where shipments slipped 10% in Q2, with only Apple gaining share, from 25% to 34% of the region’s smartphone market. In India, while phone shipments fell 11.2%, the iPhone is up 8.5%. 

Globally, the iPhone 17 was the world’s biggest-selling smartphone in Q2 2026. That means the top 10 devices are now completely dominated by Apple and Samsung. The one caveat within all this is price. Apple hasn’t yet raised iPhone prices, but is widely expected to increase them by $100 in a few weeks. When it does, the inherent value of the platform should continue to be a strong advantage for the company, which is also scooping up converts in the second-user market.

And, of course, the introduction of Apple Upgrade gives customers a route for affordable investment in new Apple kit.

What next?

Foldables appear to be the next big hope for smartphones. These expensive devices lean into the only part of the market expected to remain resilient — affluent consumers who can maintain good living standards. That’s the 25% of Americans who are interested in Apple’s new folding phone, according to Cnet. These consumers are already primed, locked, loaded and ready to buy the product, if it meets the hype.

It probably will do so, prompting IDC to predict more than 10 million sales in the first year, despite the likely $2k+ price tag and much-reported limited availability.

“A very elite set of consumers are being targeted here, who will be lining up to buy the device,” said IDC.

Siri Ai holds the key

The wild card is Siri and AI. Apple is digging in with Siri AI and is expected to introduce AI in partnership with Alibaba in its second biggest smartphone market, China, potentially alongside the new iPhones. There’s a lot riding on this, and initial reviews of what it has accomplished in Siri AI have been very positive. Apple can be confident that it now offers the very best private, potentially on-device consumer AI play, backed up with an end-to-end computing system that supports sovereign and on-prem AI services in the form of its fantastic Macs.

What does that mean? Rothschild & Co Redburn recently upgraded Apple to “buy” with a $400 target, speculating the company could become, “the gatekeeper of consumer AI, delegating AI workloads to a subservient fleet of open-source models.” 

Creative Strategies analyst Carolina Milanesi puts Apple’s AI strategy into few words: “Apple wants to own as much of the AI workflow as it can hold,” she said.

Apple’s relative smartphone market share combined with its fast-growing Mac market share means it has a compelling offer as it seeks to stake its claim in that space. What’s open to question is whether iPhone users will be comfortable using AI more frequently on their device, while Apple’s challenge is ensuring the privacy it promises is a commitment it can keep. 

You can follow me on social media! Join me on BlueSky, LinkedIn, Mastodon and subscribe to my excellent, hand-curated daily Apple news headline summary at The Core.

Cybersecurity researchers have discovered a cluster of 18 Google Chrome and one Microsoft Edge extensions that were published over the last six months and harbored wallet secret stealing and cryptocurrency draining capabilities. The extensions, per Socket security researcher Karlo Zanki, share similarities in code and tradecraft, with evidence indicating that the campaign may have been active

Cybersecurity researchers have discovered a cluster of 18 Google Chrome and one Microsoft Edge extensions that were published over the last six months and harbored wallet secret stealing and cryptocurrency draining capabilities. The extensions, per Socket security researcher Karlo Zanki, share similarities in code and tradecraft, with evidence indicating that the campaign may have been active Ravie Lakshmananhttp://www.blogger.com/profile/[email protected]