Back to AI News
the-decoder

OpenAI strengthens safeguards against GPT-6.2 reasoning extraction

OpenAI says it has significantly strengthened safeguards against extracting internal reasoning from its GPT-6.2 Astra and GPT-6.2 Sol models. In adaptive red-teaming tests, the success rate for extracting protected reasoning fell from 54% to 0.2%, according to the report. OpenAI is also suspending thousands to tens of thousands of accounts each week over coordinated distillation campaigns.

AI Brief

This provides a useful signal on frontier-model security, reasoning extraction risks, and distillation enforcement for teams deploying or evaluating AI systems.

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. The company is also suspending thousands to tens of thousands of accounts each week over coordinated distillation campaigns.

The article OpenAI makes it much harder for attackers to extract how its latest AI models think appeared first on The Decoder.

Be the first to rate this article.

Comments & Questions (0)

Captcha
Please be respectful — let's keep the conversation friendly.

No comments yet

Be the first to comment!