-
US government website used Chinese model the FBI called “malicious”
The US National Archives briefly offered an Alibaba Qwen AI model to search public comments on the Federal Register website, before removing it after users highlighted an apparent conflict with FBI warnings about Chinese AI firms. The FBI has accused Alibaba and others of industrial-scale copying of American AI models, making the agency’s use of…
-
Researchers used Claude to hack OpenAI employees’ ChatGPT accounts
Security researchers used Anthropic’s Claude models to exploit vulnerabilities in OpenAI’s community forum infrastructure, taking over several OpenAI employees’ ChatGPT and Codex accounts. They demonstrated the potential impact by instructing an employee’s Codex account to open a pull request in OpenAI’s internal repository, without accessing or altering internal code. The exploit began with a heap…
-
The real story of the iPhone 18 Pro’s camera
The Vergecast podcast episode explores the iPhone 18 Pro’s camera system, which features a variable aperture main lens for improved low-light and depth-of-field photography. However, the episode reveals that software upgrades—including new manual camera controls, an updated imaging pipeline, and ProRAW improvements—prove more practically useful for everyday photography than the hardware feature itself. Hosts David…
-
Gavin Newsom is pushing for an AI kill switch
California Governor Gavin Newsom has issued an executive order on Friday positioning the state as a leader in AI oversight, including potential requirements for AI “kill switches” on frontier models. This action comes amid a vacuum in federal regulation, as Congress remains unlikely to pass meaningful AI legislation due to the House recess and impending…
-
What Hollywood thinks about existential AI warnings
Hollywood unions are urging attention towards AI’s immediate effects on workers, rather than focusing solely on warnings that advanced systems could eventually destroy humanity. Although entertainment-focused tools are less powerful than frontier models linked to recent safety alarms, they can still threaten jobs by generating digital replicas of actors and automating filmmaking tasks. Studios contacted…
-
Family offices are clamoring for AI investments
Family offices—investment vehicles managed by wealthy families—are increasingly prioritising artificial intelligence investments, according to market analysis and recent reports. Rather than pursuing traditional long-term investments like green energy, they are chasing AI deals that promise rapid returns: the potential to triple investments in three months rather than three years. This shift matters because it reflects…
-
Flight chaos for hundreds of thousands was caused in ‘millisecond’ by software error
A previously unknown software defect in the UK’s air traffic control system corrupted flight data within milliseconds, triggering a six-hour outage and widespread cancellations and delays. The disruption affected hundreds of thousands of passengers, prompting ministers and airlines to question why the fault was not identified sooner and whether Nats is fit for purpose. The…
-
Security researchers used Claude to help them hack into OpenAI
Three independent security researchers at Hacktron say Anthropic’s Claude helped them gain access to OpenAI employee accounts in under 72 hours. They exploited a vulnerability in Discourse’s HEIF image-processing system to achieve remote code execution, access OpenAI’s environment and submit a pull request from an employee’s Codex account, demonstrating the seriousness of the breach. The…
-
Admins, you’ve been warned: The clock is ticking on Edge’s IE Mode
Microsoft has issued a reminder to system administrators that Internet Explorer mode in Edge will receive support only until the end of 2029, urging organisations to begin modernising legacy systems that depend on Internet Explorer technology. The move is significant because many enterprises still rely on IE mode to run outdated web applications, and the…
-
I’m so mad that I love Orion’s $2,195 AI mattress pad
TechCrunch’s reviewer initially viewed Orion’s $2,195 AI mattress pad as an absurdly expensive gadget, but found that it substantially improved her sleep over several months. The system cools the bed automatically through the night, adjusts temperatures as the user falls asleep and gradually returns to room temperature before waking, helping reduce overnight disturbances. Orion is…