OpenAI says new safeguards cut reasoning-extraction attacks

OpenAI said on Nov. 20 that updated safeguards on GPT-6.2 Astra and Sol tie protected reasoning to the conversation that produced it and block older model families from reading those traces. In the company's own adaptive test, the average success rate for extracting reasoning from an earlier request fell from 54.0 percent on GPT-6 Astra to 0.2 percent on GPT-6.2 Astra, and from 22.9 percent to 0.9 percent for the current request. OpenAI said it is applying selected improvements to GPT-6 Astra as well. The figures are internal and have not been independently reproduced.

/ Sources

/ About this story

Compiled by Venture Atlas from the sources above, using automated AI-assisted research. This is a summary of reporting published elsewhere, not original reporting - follow the source links for the full account. See our editorial standards.

Something wrong here? Email apostructura@gmail.com and we will fix it.

/ Related