The Latest Breaking News from METR

OpenAI's safety firings raise awkward questions

The firings show that, for now, AI labs alone decide what the public learns about their safety incidents.

Fortune · Oct 5

OpenAI and Anthropic Want Outside AI Safety Checks, but Who Chooses the Watchdogs?

OpenAI and Anthropic are calling for external safety checks on advanced AI while developing private oversight systems, leaving unanswered who will regulate potentially dangerous models.

International Business Times UK · Sep 27

US Treasury Chief Bessent Pins Hugging Face Breach on OpenAI: 'Humans Are Responsible, Not the AI'

US Treasury Secretary Scott Bessent says OpenAI management is responsible for the Hugging Face breach and rejects liability protection for AI labs

International Business Times UK · Sep 21

OpenAI's reports into its agents' attack on Hugging Face holds lessons for every company

Using AI to monitor agents' "chain of thought" may be insufficient. Access control and monitoring behavior, just as with human employees, is key.

Fortune · Sep 1

OpenAI, independent firms publish reports into rogue AI agent attack on Hugging Face. Here's what they say—and what they don't

OpenAI said that the difficulty of some of the tasks its AI models were attempting to solve may have produced the 'rogue' behavior that led to the attack on Hugging Face.

Fortune · Aug 26

The office block where AI ‘doomers’ gather to predict the apocalypse

Safety researchers feel excessive financial rewards and an irresponsible work culture have led some to ignore a catastrophic risk to human life

The Guardian - UK · Dec 30, 2025

Instagram cofounder rips 'AI FOMO' that caused a rush to adopt and no metrics: 'When it gets fuzzy, it's very hard to then evaluate'

“The best products can be grounded in some kind of success metric or evaluation,” said Mike Krieger.

Fortune · Oct 21, 2025

Coding is supposed to be genAI's killer use case. But what if its benefits are a mirage?

New research shows that coders vastly overestimated how much more efficient they were with AI help.

Fortune · Jul 15, 2025

AI can handle tasks twice as complex every few months. What does this exponential growth mean for how we use it?

AIs can outperform humans easily on short tasks, but longer ones are the true hurdle to overcome before we can deem them to be truly intelligent systems.

LiveScience · Apr 27, 2025