OpenAI has strengthened safeguards on its latest models, GPT-6.2 Astra and GPT-6.2 Sol, cutting the success rate of attempts to extract protected reasoning from 54 percent to 0.2 percent, and is suspending large numbers of accounts tied to coordinated distillation campaigns.
OpenAI sharply cuts attackers' ability to extract its AI models' reasoning
OpenAI has strengthened safeguards on its latest models, GPT-6.2 Astra and GPT-6.2 Sol, cutting the success rate of attempts to extract protected reasoning from 54 percent to 0.2 percent, and is suspending large numbers of accounts tied to coordinated distillation campaigns.
Monitoring item. The full text is not distributed. Extract and source below.