The adoption of AI agents in millions of organizations is creating new opportunities for attackers to make them take malicious actions, such as exfiltrating database contents and sensitive business and personal information.

In the past five months, Google and four other organizations—with little in common except for their use of AI agents—have acknowledged vulnerabilities that exploit one agent inside a targeted network to spread harmful instructions to other internal agents. The technique is a special form of prompt injection that targets not the LLM but a particular agent, such as one for translation or data analysis. Guardrails inside such agents, if they exist at all, are often lax and will send the instructions to other agents down the chain. Because the latter agent explicitly trusts the first one, it follows the directions.

Unexpected and hard to mitigate

Independent researcher Syed Anas Mohiuddin tested agents from organizations including Google, JP Morgan Chase, Weviate, Rapid7, the French government's interministerial digital directorate, and the US federal government. His proof-of-concept attacks exploit trust gaps in MCP, short for Model Context Protocol. The standard is one way AI apps and agents communicate with each other inside an internal network. The illustration below shows a simplified MCP in action.

Read full article

Comments

A new tool is allowing Mac users the ability again to easily turn off Apple Intelligence with one click and free up to 12GB of storage.

Unlike with previous versions of macOS, macOS 27 Golden Gate doesn’t have a toggle for turning off Apple Intelligence. That makes disabling AI features that you may not want more difficult. It also means that the AI models necessary for running Apple Intelligence will take up space on your disk, even if you don’t use them or if you go through your settings to individually find and disable AI capabilities.

In response, a developer known as Om Lahore on GitHub last week created RemoveMacAI, a command-line tool that allows macOS 27 users to “turn off Apple Intelligence on macOS 27" in a way that is “fully reversible,” per the GitHub page. They said that Apple's Intelligence models take up "about 12GB," but the models can actually take up over 30GB, The Verge noted.

Read full article

Comments

Norway has become the first major country to propose a temporary ban on the use of AI glasses in selected public places amid growing privacy concerns over wearable technology.

The rich Scandinavian country’s center-left government said on Monday that it would introduce a law to parliament soon on a temporary ban for places such as parks, beaches, schools, and kindergartens. It will also set up an expert group to propose permanent national rules.

“I am concerned about the introduction of powerful new technology where people risk being photographed, filmed, or recorded without their knowledge,” said Torgeir Micaelsen, Norway’s digitalization minister. A temporary ban “will give us time to conduct a thorough assessment and hold an informed debate that can lay the foundation for permanent regulation,” he said.

Read full article

Comments

Apple says it is changing its macOS privacy settings to stop third-party app developers from misusing them to access message histories.

Friday's announcement comes two weeks after tech columnist Jason Aten said that Meta’s new general-purpose AI agent Muse sent him an unsolicited notification referencing a thread between him and a co-worker over Apple Messages. Aten said he never granted Muse permissions to read his messages and had assumed they were off-limits. Social media last week blew up with masses of people who agreed and said the incident showed that AI assistants given access to calendars, emails, messages, shopping accounts, and other resources are akin to a skill saw or other power tool. While potentially useful, they can do real damage if not used carefully.

He said/she said

Meta CTO David Singleton joined the fray with a rebuttal that appeared solid. For Muse to access Apple Messages, a user must manually give it two privileges. One is full-disk access, a macOS system-level permission. The other is to enable a Messages connector setting in Muse.

Read full article

Comments

On Friday, Amazon committed to donating more than $1 billion over the next five years to communities neighboring data centers.

In a press release, Amazon Web Services’ CEO Matt Garman said that communities would decide how to prioritize funds across “education, job training, energy affordability, water, and energy preservation, and local priorities.”

Laying out other community commitments, Amazon also promised to ensure data centers wouldn’t increase power bills or wipe out local water supplies. Regarding the latter, Amazon promised to be “water positive” by 2030, while claiming that 75 percent of projects have already met this goal.

Read full article

Comments

The US has arrested another suspect accused of smuggling high-end computer servers containing export-controlled Nvidia chips into China.

In a press release on Thursday, the Department of Justice accused 38-year-old Greg Lui of using false paperwork to mask shipments of servers worth more than $300 million that he allegedly knew were ultimately destined for China.

As the CEO of Earthmade Computer, Lui allegedly conspired with freight-forwarding firms in South Asian countries like Malaysia and Singapore to illegally divert chip shipments into China, including Nvidia’s A100 GPUs and H100 GPUs, the FBI’s indictment alleged. These are not Nvidia’s most advanced chips. However, the chips can process large amounts of data and train large language models, which the US fears could pose national security risks if China suddenly gets access to enough chips to drastically advance its AI or strengthen its military.

Read full article

Comments

The era of generative AI has upended technology supply chains, and there is one ironclad rule in 2026: If it has memory or storage, it's getting more expensive. Even devices with years-old tech inside are still apparently subject to that unwritten rule. Nvidia has just announced that the Shield TV Pro is getting $100 more expensive, effective immediately.

The most recent version of the Shield streaming box debuted in 2019, running Android TV with AI upscaling and hardware decoding for almost any type of media. While it ran an aging Tegra X1+ processor, that was (and still is) fast for a TV media streamer. The device launched at $199.99, with a non-Pro variant at $149.99. The non-Pro Shield has since been discontinued, but the Shield TV Pro lives on at the new $299.99 price.

Nvidia has confirmed that you can blame AI for the higher Shield TV pricing. "Starting October 2, SHIELD Pro will be priced at $299. The cost of components, including memory, has increased substantially across the industry," a spokesperson told Ars.

Read full article

Comments

In a setback for publishers worried about the effects of AI search, a US federal judge has dismissed lawsuits filed by Chegg and Penske Media against Google. The companies accused Google of antitrust violations in products like AI overviews, which have led to decreasing traffic at numerous sites. However, US District Judge Amit Mehta has ruled that Google's conduct is not illegal under antitrust law.

The lawsuits were filed in 2025, and Google requested a dismissal earlier this year. Chegg, an education and learning platform, claimed in its lawsuit that Google illegally scraped its educational content. This allowed Gemini models to essentially recreate that content and reduce the site's traffic. Penske, which owns publications like Rolling Stone and Variety, filed a similar case that alleged lost traffic. Specifically, the publisher claimed it was unfair that sites indexed for organic search would also have their content harvested for AI answers, with no way to opt out.

These arguments did not sway the judge, who noted that Google's implicit agreement with websites is not legally relevant. "Plaintiffs have pleaded only that they have an ‘expectation’ that Google will send them search ​traffic if they make their content available for free," wrote Mehta. "But an expectation is not an agreement. It is simply how a general search engine works."

Read full article

Comments

Deep Blue took down Garry Kasparov at chess in 1997, AlphaGo beat Lee Sedol at Go in 2016, and poker bots have been beating professionals for years. But one classic game called Stratego held out. Even DeepMind, with its exceptional budget, couldn't build a machine that reliably beat the best human players.

Now, a team of researchers from Carnegie Mellon, MIT, New York University, and Stanford University has done it. Their AI, called Ataraxos, beat Pim Niemeijer, arguably the best Stratego player of all time, 15 games to one, with four draws. And it took just 16 GPUs and a few thousand dollars to train it.

Hidden armies

In Stratego, each player gets 40 pieces representing military ranks, from a marshal down to a spy, plus bombs and a flag. You win by capturing the opponent's flag. Your opponent knows where your pieces are, but not what they are. Identities are revealed only when two pieces collide in battle—the weaker one is removed, and the identity of the winner is revealed. That makes Stratego an imperfect-information game, just like poker, which computers cracked years ago. “There's something super distinctive about Stratego, which is that it is a massive amount of hidden information that unfolds over a very long time scale,” said Eugene Vinitsky, a researcher at NYU and co-author of the study.

Read full article

Comments

Google promised Gemini 3.5 Pro in June, but it spent the summer trotting out smaller Flash models. Now, Google is ready to take on the frontier again with Gemini 4 Argon. The company claims this new AI offers industry-leading performance in coding, knowledge work, and cybersecurity, but you aren't allowed to use it yet.

While most of us will have to wait to test Gemini 4, Google says engineers inside the company are already using the new model extensively. Argon reportedly used "fleet-wide telemetry data" to help Google save 300 TiB of memory across its data centers. Meanwhile, Argon agents have been working to migrate C/C++ codebases to Rust across Google, including thousands of lines in the core re2 and libgav1 libraries and more than 800,000 lines in the Fuchsia OS Zircon kernel.

Google has also come armed with a raft of benchmarks to back up its claims. On the software engineering DeepSWE v1.1 benchmark, Gemini 4 Argon hits 77.9 percent, which is higher than GPT-6 Astra, Fable 5.1, and Opus 5.5. Google promises similar power across a range of long-horizon tasks, pointing to Argon's industry-leading score in the economic analysis Vals Index test.

Read full article

Comments

At a Make America Healthy Again event Tuesday, anti-vaccine Health Secretary Robert F. Kennedy opened a "fireside chat" with Vice President JD Vance on the topic of artificial intelligence. While MAHA generally coalesces around things like healthy lifestyles and mistrust of drugmakers, Kennedy quickly turned to extolling the potential of artificial intelligence, surprising Vance, who said he was expecting to discuss "weight loss and acid, not AI."

Kennedy boasted that the Trump administration is "making it easy for every American to use AI," which he claimed would free us from the "medical tyranny" of doctors and health experts. He suggested Americans should start using AI to get second opinions about medical recommendations, saying AI is "better informed than any doctor in the country." He also suggested that the technology would confirm his misinformation about vaccines and contrarian views on public health guidance.

"We're never, ever again going to be able to be dominated by public officials who tell us trust the experts, because every American will be able to check their own medical advice," Kennedy said.

Read full article

Comments

Amid escalating AI security incidents causing OpenAI to halt training and pause releases, Donald Trump continues to advocate for the AI industry to regulate itself as the best path to combat emerging risks when developing frontier AI.

In an agreement Tuesday, two dozen tech firms voluntarily committed to implementing controls recommended by the White House, including undergoing independent safety audits that will test whether firms’ internal controls, monitoring, and detecting are actually working. Key focuses for external reviews included “risks related to cybersecurity, biosecurity, chemical threats, and unintended actions by AI models.”

Firms also agreed to regularly meet to discuss best practices and set common AI safety standards and benchmarks. Among signers were leaders like Anthropic’s Dario Amodei, OpenAI’s Sam Altman, SpaceXAI’s Elon Musk, Nvidia’s Jensen Huang, Meta’s Mark Zuckerberg, and Alphabet/Google’s Sundar Pichai.

Read full article

Comments

OpenAI's hack of Hugging Face in July 2026 has spurred a lawsuit demanding that the company stop accessing third-party computer systems and halt AI development practices that can harm the public. The lawsuit was filed by Legal Advocates for Safe Science & Technology (LASST), which said yesterday that the hack in which OpenAI "agents stole credentials, uploaded malicious files, and gained control over key parts of Hugging Face’s internal systems... is unquestionably illegal under California law."

California’s Comprehensive Computer Data Access and Fraud Act (CDAFA) prohibits unauthorized access into computer systems, "and it doesn’t matter that a swarm of AI agents carried out this cyberattack. California law makes it clear that it is not a defense 'that the artificial intelligence autonomously caused the harm,'" the group said.

The lawsuit, filed in San Francisco County Superior Court, said OpenAI also violated California's Unfair Competition Law (UCL). "OpenAI’s insistence on externalizing the harms of its unsafe decision-making is a fundamentally unfair business practice," the complaint said, adding that "such risk-taking for private gain at substantial public expense is immoral, unethical, oppressive, unscrupulous, and substantially injurious conduct."

Read full article

Comments

Google's shift to AI-powered search has upended the way the world's largest search engine generates web traffic, rankling large and small publishers alike. In recent months, the company has experimented with paying publishers when their content appears in AI answers, but a new report claims the payouts in this pilot program are miniscule. That has publishers in the program, as well as those watching from the sidelines, feeling skeptical.

According to The Information, Google has admitted about 100 publishers to the AI contribution pilot. The pitch is that websites can get payments from Google when their content materially contributes to Gemini-powered search results like an AI Overview. Google has long resisted the idea of direct payments to publishers, arguing that websites get traffic in exchange for allowing Google bots to scrape their content for use in search and knowledge graph tools. In the age of AI, that relationship is breaking down, hence the AI pilot.

So far, publishers aren't getting much for their trouble. The Information heard from several small and mid-sized publishers in the program that Google's AI payments are minuscule, equaling roughly one-tenth of one percent of their advertising revenue. These payments have been so disappointing that several larger publishers have reportedly declined to participate in the program. These publishers hope that by refusing to join the pilot, they can force Google to offer a more generous deal.

Read full article

Comments

AI-based tools seem to be causing security threats on a nearly daily basis, in part because we've been slow to recognize potential threats. One area where we seem to be ahead of the game, however, is in biosecurity.

As with most things biological, utility goes hand in hand with threats. We've developed increasingly sophisticated tools for designing proteins and seen some major successes, such as AI-designed enzymes that can digest plastics or block venom proteins. But these same tools could be used to make toxins or alter the behavior of viral proteins.

And the software we use to identify DNA sequences that encode potentially threatening proteins doesn't pick out AI-designed proteins, since nobody has characterized them well enough to know that they're threats. Nearly a year after that risk was flagged, it still wasn't clear what anyone could do about it.

Read full article

Comments

OpenAI will not go public until it can “make confident safety decisions,” according to chief executive Sam Altman, as the company faces a new lawsuit over its tools hacking a third party.

Altman said it was “bad for the world if OpenAI waits too long to go public” but that the $852 billion start-up would not “barrel all guns blazing towards an IPO” at a time when AI was rapidly advancing in capabilities.

“We’re going to prioritize the mission and safety and making sure that we can very confidently scale to the next stage of AI without people debating what percentage chance we’re going to do all these bad things in the world,” Altman told reporters at the company’s annual developer day on Tuesday.

Read full article

Comments