Introducing Trusted Access for Cyber
OpenAI is launching Trusted Access for Cyber, a trust-based framework designed to give cybersecurity defenders priority access to its most capable frontier models while implementing safeguards against misuse. This approach addresses a core challenge: distinguishing between legitimate defensive work (patching vulnerabilities) and malicious activity (exploiting them).
How It Works
The framework operates on two tracks:
- Individual users can verify their identity at chatgpt.com/cyber to access enhanced cyber capabilities
- Enterprise teams can request trusted access through their OpenAI representative for organization-wide default access
Security researchers and teams needing even more permissive models for advanced defensive work can apply to an invite-only program. All users with trusted access must comply with OpenAI's Usage Policies and Terms of Use.
Safety and Safeguards
GPT-5.3-Codex includes built-in mitigations trained to refuse clearly malicious requests like credential theft. The model is paired with automated classifier-based monitors to detect suspicious cyber activity. OpenAI expects to refine these policies and classifiers over time based on feedback from early participants. Prohibited activities include data exfiltration, malware creation or deployment, and unauthorized testing.
Financial Support for Defenders
OpenAI is committing $10 million in API credits through its expanded Cybersecurity Grant Program. The initiative targets teams with proven track records identifying and remediating vulnerabilities in open source software and critical infrastructure. Interested teams can apply through OpenAI's official grants form.
Context
As frontier models grow more capable of autonomous, long-duration tasks, the potential for both defensive and malicious use increases. OpenAI positions this initiative as critical infrastructure for the broader ecosystem—enabling faster security improvements across software while reducing friction for legitimate cybersecurity work.