OpenAI Expands Trusted Access for Cyber and Launches GPT 5.4 Cyber
Summary
OpenAI says it is expanding Trusted Access for Cyber and rolling out GPT 5.4 Cyber to vetted defenders. The program uses identity verification and phased access to provide defensive capabilities while managing dual-use risks.
Key points
- OpenAI aims to expand Trusted Access for Cyber to thousands of verified individual defenders and hundreds of security teams.
- New access tiers use identity verification and trust signals; enterprises can request access for their teams.
- GPT 5.4 Cyber is described as fine-tuned for defensive tasks including binary reverse engineering, vulnerability analysis, and malware investigation.
- Access to GPT 5.4 Cyber is limited and being phased in for vetted vendors, organizations, and researchers because of dual-use concerns.
- OpenAI says its Codex Security tool has contributed to fixing thousands of high- and critical-severity vulnerabilities.
- OpenAI says some limitations will remain where usage monitoring is restricted, including zero-data-retention and certain third-party environments.
Article Details
- Event Type
- Expansion of a cybersecurity access program and launch of a specialized AI model
- Impact
- OpenAI says it is expanding Trusted Access for Cyber to serve thousands of verified individual defenders and hundreds of security teams, with access tiers based on identity verification and trust signals. GPT 5.4 Cyber is being rolled out in phases to vetted vendors, organizations, and researchers for defensive work including binary reverse engineering, vulnerability analysis, and malware investigation. OpenAI also says Codex Security has contributed to fixing thousands of high- and critical-severity vulnerabilities.