25 Fields Medalists tell AI labs to stop treating math like a benchmark

Good morning 👋 Twenty-five Fields Medalists just told AI labs, in one voice, to knock it off. When mathematicians who can't agree on lunch sign the same letter, that's the story.

In today's issue:

• 🔭 25 Fields Medalists say AI's benchmark-math race is corroding the field
• 🏛️ OpenAI's policy chief wants safety rules locked in before someone else's draft wins
• 🔗 Nvidia eyes an Anthropic IPO anchor, plus a free-for-feds ChatGPT deal and GPT-6 Astra for Wall Street

🔭 THE ONE THING

🧮 25 Fields Medalists tell AI labs to stop treating math like a benchmark

Twenty-five Fields Medalists, spanning Pierre Deligne's 1978 medal to Yu Deng's this year, signed a declaration posted yesterday on Terence Tao's blog, arguing that labs racing to solve famous open problems as benchmarks are publishing "in a rush, leaving no time for a proper writeup, the isolation of new methods and ideas, and citing relevant previous work." They admit outright they skipped the usual consultative drafting process because the moment felt too urgent to wait. Getting 25 Fields Medalists to agree on where to get lunch is hard, so a unanimous statement like this is itself the signal. Their complaint is about incentives: a lab chasing a "solved it" headline has no reason to wait for peer review, and a proof nobody can check or build on is a trophy, not math. Mathematicians rarely move this fast as a bloc. If benchmark culture keeps rewarding "solved" over "understood," this won't be the last group to push back.

🏛️ POLICY & REGULATION

• OpenAI Chief lobbyist Chris Lehane published an op-ed arguing the "policy window" for durable AI safety rules is open now and closing. Read the fine print: no bill, no filing, no agency action behind it. This is OpenAI telling Washington what it wants written into law before someone else's draft does, dressed up as a safety plea. Worth watching for what specific standards it pushes next, not worth treating as policy yet. The AI policy window is open

🔗 QUICK LINKS

• Anthropic is talking to Nvidia about anchoring its IPO, up to $10B on a raise that could hit $100B at a $2T valuation. Reuters
• OpenAI cut a deal with the GSA: $0 license fees and 50% off usage for federal, state, local, and tribal governments, plus expanded cyber defense support. OpenAI
• OpenAI launched ChatGPT for Financial Services, bundling market data with GPT-6 Astra for research, modeling, and client-ready materials. OpenAI

See you tomorrow.

Pradeep Perugu