OpenAI launches GPT-5.4-Cyber, expands defender access
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- OpenAI unveiled GPT-5.4-Cyber, a variant of GPT-5.4 optimized for defensive cybersecurity, days after Anthropic debuted its frontier model Mythos.
- OpenAI is expanding its Trusted Access for Cyber (TAC) program to thousands of authenticated individual defenders and hundreds of teams responsible for securing critical software.
- OpenAI acknowledged the dual-use risk, noting adversaries could invert defensive models to detect and exploit vulnerabilities before they can be patched.
- Codex Security, OpenAI's AI-powered application security agent, has contributed to over 3,000 critical and high fixed vulnerabilities, per the company.
- Anthropic's Mythos model, deployed via Project Glasswing in a controlled rollout, found "thousands" of vulnerabilities in operating systems, web browsers, and other software.
Why it matters: OpenAI's Trusted Access for Cyber expansion to thousands of defenders and hundreds of teams comes as Codex Security claims 3,000+ critical and high vulnerabilities fixed, while Anthropic's Mythos found "thousands" more — turning AI-assisted vulnerability discovery into a direct race between the two largest labs.
Ask SkimNews
