95.3K views · 644 likes · 47 reposts (at collection)
OpenAI Partners on a Framework for Rating Cyber Jailbreaks
- Who
- OpenAI
- What
- partnered FMF and its member firms on a framework for identifying and rating cyber jailbreaks that could enable serious cyber harm
- Quote
We partnered with @fmf_org and its member firms on a new framework to help identify when bypassing an AI system's safeguards could enable serious cyber harm, and prioritize stronger protections.
What happened
OpenAI partnered with the Frontier Model Forum and its member firms on a framework to identify when bypassing AI safeguards could enable serious cyber harm and help prioritize stronger protections. OpenAI’s Sasha Baker says it aims to provide a common methodology for comparing cyber jailbreak severity, including the capabilities they unlock and how readily they can be used.
Read next
See the full storyOpenAI and Frontier Model Forum Set a Framework for Rating Cyber JailbreaksWe partnered with @fmf_org and its member firms on a new framework to help identify when bypassing an AI system's safeguards could enable serious cyber harm, and prioritize stronger protections.
"...This framework is an important step toward a common methodology for evaluating and comparing the severity of cyber jailbreaks, including the capabilities they unlock and how readily they can be used. Shared approaches like this can help the AI and security communities build a clearer evidence base and prioritize improvements to safeguards." - OpenAI's Sasha Baker, Head of National Security Policy
The story so far
OpenAI partnered with the Frontier Model Forum and its member firms on a framework to identify and rate AI safeguard bypasses that could enable serious cyber harm, helping prioritize stronger protections. OpenAI’s Sasha Baker said the method compares the severity of cyber jailbreaks by assessing the capabilities they unlock and how readily those capabilities can be used.
Keep up with AI, one email a day
Model releases, price changes and compute developments, with original sources. Free, confirmed by email, and easy to unsubscribe. Read a past issue · 2026-11-07