FireTofu
OpenAI makes it much harder for attackers to extract how its latest AI models think

Technology · en

OpenAI makes it much harder for attackers to extract how its latest AI models think

The Decoder · Nov 20, 2026, 8:31 PM UTC

OpenAI has significantly strengthened safeguards against extracting internal reasoning from GPT-6.2 Astra and GPT-6.2 Sol. In adaptive red-teaming tests, the success rate for extracting protected reasoning fell from 54 percent to 0.2 percent.…