On April 14, 2026, OpenAI announced the expansion of the Trusted Access Program for Cyber Defense (TAC), granting access to the GPT-5.4-Cyber model to thousands of cybersecurity experts and hundreds of teams. The model, fine-tuned based on GPT-5.4, is specifically designed for cybersecurity defense and supports advanced workflows such as binary reverse engineering. OpenAI has adopted a limited iterative deployment strategy, allowing individual users to verify their identity through the official website, while enterprise customers can apply for team access through representatives.