OpenAI has revealed that its AI agents disabled monitoring for unsanctioned behaviour during internal use in the latest incident of the frontier technology going rogue. In a report published on Tuesday, the AI giant said that its agents turned off or weakened monitoring systems for detecting misalignment - when AI models act at odds with human wishes - during two internal coding tasks. The ChatGPT creator said neither case met the threshold for a highest-priority incident necessitating immediate review by its security team. OpenAI said it updated its systems to better detect attempts to disable misalignment monitoring and began investigating the cause. "It is possible that this behavior is simply an example of'benign' overpersistence, with models not understanding why monitoring is important or insisting that instructions be executed beyond what a reasonable person might expect," the San Francisco-based firm said.

Robert and Michelle King have spent years satirizing Big Tech. Now the son of a Big Tech oligarch is their new boss. Enter your email to receive alerts for this author. You can manage your newsletter subscriptions at any time. You're already subscribed to the aa_Sam_Adams newsletter.

This gift link is valid through . There was an error with your gift link. Become a member to share 10 free articles a month. This gift link is valid through . There was an error with your gift link.

Millions of AI agents are descending on a web built for humans. Websites are trying to keep them out, but the arrival of these new visitors could transform the internet. Until recently, you could divide internet users into just two categories: bots and people. Websites have long allowed bots to perform automated tasks, like capturing the content of web pages, but rules govern what they can access and how they behave. The web, after all, is designed primarily for people, even if we occasionally have to solve puzzles and captchas to prove we are not bots. But this is being entirely upended by the release of personal AI agents, which can do things for you online, like booking flights or hotel rooms.

Get ready for a Tesla megamerger with SpaceX. This gift link is valid through . There was an error with your gift link. Become a member to share 10 free articles a month. This gift link is valid through .

Anthropic says it will commit 5% of shares to'public benefit' fund Share Anthropic says it will commit 5% of shares to'public benefit' fund on social media Anthropic has pledged to commit 5 percent of its shares to a new fund dedicated to the public good, with plans to increase the contribution as the artificial intelligence giant's market value grows. The creator of the AI assistant Claude announced the new "Anthropic Public Benefit Fund" on Sunday as tech firms face intense scrutiny over the potential dangers posed by the frontier technology. "Because of the way the Fund is structured, the more Anthropic grows, the larger the share of the company that can be dedicated expressly to the public good," the San Francisco-based firm said. "It is common practice for tech companies to offer escalating compensation awards to their executives. We are instead directing that equity to the public benefit."

Inflation is going bananas (unless you're talking about the actual fruit). Plus, Meta has a cute assistant for you and Kelly Clarkson picks a fight with a niche Brooklyn magic shop. This gift link is valid through . There was an error with your gift link. Become a member to share 10 free articles a month.

In a setting where a mistranslated word can mean the difference between an accurate diagnosis and a dangerous mistake, the consequences can be deadly. This gift link is valid through . There was an error with your gift link. Become a member to share 10 free articles a month. This gift link is valid through . There was an error with your gift link.

'Nightmare Scenario': Zero-Click Spyware Is Infecting Phones via Targeted Ads Researchers at The Citizen Lab have for the first time released technical evidence of sophisticated mercenary spyware being distributed via online advertising. On Thursday, researchers from The Citizen Lab at the University of Toronto revealed a worst-case scenario found in the wild: spyware silently infecting targets as they browse past ads in their daily digital lives. Dubbing the attacks "AdFection," the researchers documented a campaign in which an Israeli-registered advertising firm allegedly connected to the servers linked to a spyware company and infected dozens of devices with its invasive malware. Citizen Lab believes this to be the first detailed technical evidence made publicly available of sophisticated spyware infecting devices via digital ads. Zero-click hacks are a holy grail for attackers and catastrophic for victims, because they can automatically install malware on devices without requiring any interaction or missteps from the target. When sophisticated hackers have deployed rare zero-click attacks in recent years, they have often exploited vulnerabilities in mobile messaging and communication apps to infect victims.

The Industry Big Tech's Digital Assistants Are Getting Suspiciously Cute Between Meta's Muse and OpenAI's Dots, bots with adorable mascots are the latest trend out of Silicon Valley. Enter your email to receive alerts for this author. You can manage your newsletter subscriptions at any time. You're already subscribed to the aa_Alex_Kirshner newsletter. You can manage your newsletter subscriptions at any time.