On Tuesday, OpenAI introduced GPT-5.4-Cyber, a specialized version of its newest artificial intelligence model designed specifically for cybersecurity defense applications. The launch comes exactly one week following competitor Anthropic’s reveal of its advanced AI system called Mythos on April 7.
Anthropic’s Mythos model is currently being used through “Project Glasswing,” a carefully managed program that allows selected organizations to test the unreleased Claude Mythos Preview system for cybersecurity protection. The model has successfully identified “thousands” of significant security flaws in operating systems, web browsers and various software applications.
The ChatGPT creator announced that GPT-5.4-Cyber will launch with restricted access, available only to approved security companies, organizations and researchers due to its more flexible design capabilities.
Additionally, OpenAI is broadening its Trusted Access for Cyber initiative to include thousands of verified individual security professionals and hundreds of teams responsible for safeguarding essential software systems, according to a company website announcement.
The tech company is introducing additional levels to its TAC program, which debuted in February, where increased verification requirements provide access to more advanced features.
Security professionals who qualify for the program’s top tier will receive access to GPT-5.4-Cyber, which operates with reduced limitations on sensitive cybersecurity operations including vulnerability research and security analysis.
