OpenAI unveils GPT-5.4-Cyber to strengthen defensive cybersecurity capabilities

The new model will be rolled out selectively to verified security professionals, as OpenAI expands its trusted access programme and intensifies focus on responsible deployment of advanced AI tools.

OpenAI has announced the launch of GPT-5.4-Cyber, a specialised version of its latest flagship artificial intelligence model designed to support defensive cybersecurity operations. The move comes shortly after rival Anthropic revealed its own frontier model, Mythos, signalling intensifying competition in AI-driven security solutions.

The newly introduced model is fine-tuned to assist cybersecurity professionals in identifying, analysing, and mitigating vulnerabilities across software systems. OpenAI said GPT-5.4-Cyber will initially be made available in a phased rollout to a select group of vetted security vendors, organisations, and researchers due to its relatively permissive operational design.

Rising competition in AI-led cybersecurity

Anthropic’s Mythos model, launched earlier this month, is being deployed under its controlled “Project Glasswing” initiative. The programme enables selected organisations to use a preview version of the model for defensive purposes. According to the company, the system has already helped uncover thousands of significant vulnerabilities across operating systems, web browsers, and other widely used software platforms.

The rapid development of such tools highlights a growing trend among AI companies to build specialised models that can proactively strengthen cybersecurity frameworks, particularly as digital threats become more sophisticated and widespread.

Expanded access through trusted programmes

Alongside the launch, OpenAI is also expanding its Trusted Access for Cyber (TAC) programme to broaden participation among verified cybersecurity professionals. The initiative, introduced earlier this year, is being scaled to include thousands of individual defenders and hundreds of teams working to secure critical infrastructure and software systems.

New access tiers are being added to the programme, with higher levels of verification granting users access to more advanced capabilities. Those approved at the highest level will be able to utilise GPT-5.4-Cyber with fewer restrictions, particularly for sensitive tasks such as vulnerability research and threat analysis.

The development reflects OpenAI’s approach to balancing innovation with responsible deployment, ensuring that powerful cybersecurity tools remain accessible to trusted stakeholders while mitigating potential misuse.