OpenAI Says GPT-6.2 Astra Cut Reasoning Extraction Success to 0.2%
5 Articles
5 Articles
OpenAI massively strengthens the protection of internal thinking processes at GPT-6.2 Astra and GPT-6.2 Sol. In adaptive red-teaming tests, the success rate for reading protected thought processes dropped from 54 to 0.2 percent. At the same time, the company blocks thousands to tens of thousands of accounts per week that are involved in coordinated distillation campaigns. The article AI-Datenklau: OpenAI makes imitators' models hard to live appe…
OpenAI makes it much harder for attackers to extract how its latest AI models think
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 A…
The company claims that its new defenses reduced the effectiveness of certain protected reasoning extraction attacks from 54% to 0.2%. It also requests measures The post OpenAI claims to reduce 99.6% attacks to extract GPT-6.2 reasoning and calls for sanctions against foreign actors appeared first on .
Coverage Details
Bias Distribution
- There is no tracked Bias information for the sources covering this story.
Factuality
To view factuality data please Upgrade to Premium





