OpenAI Launches GPT-5.4-Cyber with Expanded Access for Security Teams

Key Points:

• OpenAI introduced GPT-5.4-Cyber, a specialized version of GPT-5.4 optimized for defensive cybersecurity, aiming to help defenders identify and fix digital infrastructure problems more quickly.
• The company is expanding its Trusted Access for Cyber (TAC) program to thousands of individual defenders and hundreds of security teams responsible for critical software protection.
• OpenAI acknowledges the dual-use nature of AI, emphasizing efforts to democratize access while minimizing misuse risks, such as adversaries exploiting defense models to find vulnerabilities before patches are released.
• The rollout strategy focuses on responsible use at scale, strengthening safeguards against jailbreaks and adversarial prompts as model capabilities evolve, while broadening legitimate defender access.
• This announcement follows Anthropic's preview of its Mythos model, which has identified thousands of vulnerabilities, highlighting a broader industry trend toward integrating AI into continuous software security and risk reduction workflows.