OpenAI makes it much harder for attackers to extract how its latest AI models think
The Decoder
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.
Summary drawn from The Decoder's own description of the story.