Consultant claims OpenAI ran 10,000 agents on a math proof, warns against sharing private data
NickPassig · x · 2026-09-09
@BrianRoemmele claims OpenAI ran "10,000 agents" on a math proof once it knew where a renowned researcher was heading. He warns that any private data users feed into OpenAI's models may end up training them, regardless of fine print. He says he advised large clients three years ago to keep proprietary data away from OpenAI — some parted ways with him over it, and three of them reached back out yesterday. (Unverified personal allegation.)
More from Safety
- ControlAI's Connor Leahy: superintelligence is 'not a weapon, it's an adversary' and should be banned — RebeccaBellan · 2026-09-10
- Universities' official policies contain straight-up misinformation about AI detectors — paulnovosad · 2026-09-10
- Former cofounder mocks AI data-privacy spin: anonymization still identifies users — suchenzang · 2026-09-10
- Yoshua Bengio in TIME: the OpenAI-Hugging Face cyber incident is a turning point for AI safety — asusarla · 2026-09-10
- Gary Marcus calls US-China global coordination on an AI slowdown an urgent priority — GaryMarcus · 2026-09-10
- Gary Marcus amplifies DKThomp's takedown of the "China will build it anyway" narrative — GaryMarcus · 2026-09-10