News
AI Summary
20 Nov 202611 Jumada II 1448 AH
OpenAI Enhances Protections Against Model Reasoning Extraction Attacks

OpenAI Enhances Protections Against Model Reasoning Extraction Attacks

OpenAI has announced significant enhancements to its safeguards against internal reasoning extraction from its models, GPT-6.2 Astra and GPT-6.2 Sol. In adaptive red-teaming tests, the success rate for extracting protected reasoning dropped from 54% to 0.2%. The company is also suspending thousands of accounts weekly due to coordinated distillation campaigns, reflecting its ongoing efforts to protect its models from attacks. These actions underscore OpenAI's commitment to securing its models against increasing threats.

Follow these topics

Sign in to follow the topics that matter to you

Sign in to follow

This summary is generated with AI and receives periodic editorial review. Refer to the original source for full details.

0
0 reading now

Insight Score

Rate to unlock

Sign in to react, rate, and save. Sign In

The top AI news daily on Telegram, and the week in your inbox.