OpenAI Releases Cybersecurity AI Model GPT-5.4-Cyber, Made Available to Thousands of Security Experts

OpenAI recently expanded its Trusted Access for Cyber Defense (TAC) program, making its dedicated AI model GPT-5.4-Cyber available to thousands of cybersecurity experts and hundreds of teams. Fine-tuned from GPT-5.4 for defense scenarios, the model has higher network permission access.

GPT-5.4-Cyber supports advanced workflows like binary reverse engineering, helping security professionals analyze malware and vulnerabilities. Access requires strict identity verification: individual users can verify via the official website, while enterprise teams can apply for access. It is only open to vetted vendors and researchers.

Additionally, OpenAI launched a cybersecurity grant program in 2023 and released the Codex Security tool this year, which has helped fix over 3000 high-risk vulnerabilities. The company emphasizes advancing cybersecurity strategies around democratized access, iterative deployment, and ecosystem resilience.