aiminute. ← All AI news
Research AI Minute Newsroom 2026-11-22

OpenAI says its new models block nearly all reasoning-theft attacks.

OpenAI says its new models block nearly all reasoning-theft attacks.

OpenAI said attempts to extract its models' hidden reasoning now succeed 0.2% of the time, down from 54%. The figures come from the company's own tests on GPT-6.2 Astra and are not independently checked. It said it blocks thousands to tens of thousands of accounts each week for such copying.

Why it mattersDistillation means training a cheaper model on a stronger model's answers. OpenAI attributed one September campaign to people linked to Moonshot AI, the Chinese maker of Kimi.

✓ Verified · 1 sources

WhatsApp X Telegram
Read in the app — free, in 9 languages

Related stories

Lean's own community says the AI-assisted Riemann proof is correct.
2026-11-22
Researchers say an open AI model improved by editing its own weights.
2026-11-21
A US watchdog found that no AI therapy chatbot has FDA authorization.
2026-11-21
Outside mathematicians say the AI Riemann proof passed first checks.
2026-11-21
A test with two million bait prompts caught AI services leaking data.
2026-11-21