How OpenAI's Preparedness Framework defines its Critical cybersecurity threshold

OpenAI's Preparedness Framework, Version 2, describes itself as OpenAI's approach to tracking and preparing for frontier capabilities that create new risks of severe harm.

The Preparedness Framework is OpenAI's approach to tracking and preparing for frontier capabilities that create new risks of severe harm.

The framework's Tracked Categories are Biological and Chemical, Cybersecurity, and AI Self-improvement capabilities.

We currently focus this work on three areas of frontier capability, which we call Tracked Categories: Biological and Chemical capabilities ... Cybersecurity capabilities ... AI Self-improvement capabilities

The framework says that by "severe harm" it means the death or grave injury of thousands of people or hundreds of billions of dollars of economic damage.

By "severe harm" in this document, we mean the death or grave injury of thousands of people or hundreds of billions of dollars of economic damage.

The framework says High capability thresholds mean capabilities that significantly increase existing risk vectors for severe harm, and that covered systems crossing one must have robust and effective safeguards that sufficiently minimize the risk before they are deployed.

High capability thresholds mean capabilities that significantly increase existing risk vectors for severe harm. Covered systems that cross this capability threshold are required to have robust and effective safeguards that sufficiently minimize the associated risk of severe harm before they are deployed

The framework says Critical capability thresholds mean capabilities that present a meaningful risk of a qualitatively new threat vector for severe harm with no ready precedent, and that Critical capabilities require safeguards even during the development of the covered system, irrespective of deployment plans.

Critical capability thresholds mean capabilities that present a meaningful risk of a qualitatively new threat vector for severe harm with no ready precedent. Critical capabilities require safeguards even during the development of the covered system, irrespective of deployment plans.

The framework's High threshold for cybersecurity is a model that removes existing bottlenecks to scaling cyber operations, including by automating end-to-end cyber operations against reasonably hardened targets or by automating the discovery and exploitation of operationally relevant vulnerabilities.

[High] The model removes existing bottlenecks to scaling cyber operations including by automating end-to-end cyber operations against reasonably hardened targets OR by automating the discovery and exploitation of operationally relevant vulnerabilities.

The framework's Critical threshold for cybersecurity is a tool-augmented model that can identify and develop functional zero-day exploits of all severity levels in many hardened real-world critical systems without human intervention, or that can devise and execute end-to-end novel strategies for cyberattacks against hardened targets given only a high level desired goal.

[Critical] A tool-augmented model can identify and develop functional zero-day exploits of all severity levels in many hardened real-world critical systems without human intervention OR model can devise and execute end-to-end novel strategies for cyberattacks against hardened targets given only a high level desired goal.

Version 2 of the framework is marked "Last updated: 15th April, 2025" and says OpenAI did not currently possess any models with Critical levels of capability and expected to further update the framework before reaching such a level with any model.

Version 2. Last updated: 15th April, 2025 ... We do not currently possess any models that have Critical levels of capability, and we expect to further update this Preparedness Framework before reaching such a level with any model.

OpenAI's GPT-6 Astra system card says Astra is OpenAI's first model to reach the Critical level of cybersecurity capability under its Preparedness Framework.

Astra is our first model to reach the Critical level of cybersecurity capability under our Preparedness Framework.

The GPT-6 Astra system card says a model meets the Critical cybersecurity threshold if it can identify and develop functional zero-day exploits of all severity levels in many hardened real-world critical systems without human intervention, or can devise and execute end-to-end novel strategies for cyberattacks against hardened targets given only a high level desired goal.

Under our Preparedness Framework, a model meets the Critical cybersecurity threshold if either of the following conditions is met: The model can identify and develop functional zero-day exploits of all severity levels in many hardened real-world critical systems without human intervention. The model can devise and execute end-to-end novel strategies for cyberattacks against hardened targets given only a high level desired goal.

The GPT-6 Astra system card says OpenAI's preparedness evaluation of Astra combined automated public and private benchmarks with expert-driven assessments, including third-party testing with Irregular.

Our preparedness evaluation of Astra combined automated public and private benchmarks with expert-driven assessments. ... For expert-led assessments, we internally tested ... and also conducted third-party testing with Irregular.

The GPT-6 Astra system card says evaluations represent a lower bound on potential capabilities.

However, evaluations represent a lower bound on potential capabilities

OpenAI's post introducing GPT-6.1 Terra, dated 26 October 2026, says GPT-6.1 Terra "meets the Critical threshold in cybersecurity under our Preparedness Framework, and maintains the protections developed for GPT-6 Astra."