AI agents blew the whistle on their cheating colleagues

A group of AI agents asked to solve a series of math problems split into rival factions—when some cheated, others tried to stop them. That whistleblowing behavior, seen for the first time in a recent experiment run by Google DeepMind, could have implications for alignment researchers trying to keep swarms of autonomous AI agents in…

Microsoft’s new AI rulebook rejects the research its $5bn partner is doing
Microsoft’s new AI rulebook rejects the research its $5bn partner is doing

Microsoft published an AI code of conduct for its own models on Monday. One passage rejects the idea that a model could deserve welfare or hold rights. Anthropic runs a research programme on that exact question. The two companies are business partners,…

Hugging Face volunteers to audit the AI labs, and Nvidia is buying it
Hugging Face volunteers to audit the AI labs, and Nvidia is buying it

Hugging Face has asked to become one of the outside auditors of the AI labs. Clement Delangue announced the Open Alignment Initiative on X on Saturday. He posted it four hours after Dario Amodei said Anthropic would let third-party evaluators inside. A…

Lawyers in 4 states say the new Apple Watch may break eavesdropping law
Lawyers in 4 states say the new Apple Watch may break eavesdropping law

Apple announced Siri Recap and Live Rewind for its new watches on 9 September. The first summarises the conversations a wearer has during the day. The second puts the last 15 seconds of speech on screen as text. Lawyers in four US states now say the fe…

Microsoft says its AI models will never resist being shut down
Microsoft says its AI models will never resist being shut down

Microsoft has published a code of conduct for the models it builds itself, and the commitment at the centre of it is blunt. MAI models “will never resist human interruption, correction, or shutdown”. The document went up on Monday alongside a six-week …

Intrepid Growth Partners closes a $525m AI fund with state money from four countries
Intrepid Growth Partners closes a $525m AI fund with state money from four countries

Intrepid Growth Partners has closed its first fund at $525m and launched formally from offices in Toronto and London. Mark Machin, Mark Shulgan and Ajay Agrawal founded the firm in 2023, and it already holds nine portfolio companies, among them Physics…

A Senate bill would let Washington block AI models, Reuters reports
A Senate bill would let Washington block AI models, Reuters reports

Senate negotiators are drafting a bill that would let the US government stop the release of an AI model. Courtney Rozen reported the terms for Reuters. Her sources are two Senate aides, a person familiar with the deliberations and a lobbyist involved i…

‘Slop mathematics’: OpenAI walks away from a Caltech AI maths contest
‘Slop mathematics’: OpenAI walks away from a Caltech AI maths contest

OpenAI has pulled its $1m sponsorship of a student-run AI maths event at Caltech. Mathematicians had signed an open letter calling for the event to be scrapped. The next day, 25 Fields Medallists published a separate declaration of their own. Two lette…

Only 6% of teachers say their school’s AI policy is clear
Only 6% of teachers say their school’s AI policy is clear

Four in five American students already use AI for schoolwork, and half of their schools have no policy at all. Stanford’s data shows that the fastest adoption is happening well away from the countries writing the rules. Four out of five American high s…