OpenAI unveils cybersecurity model with limited rollout days after Anthropic’s model

OpenAI released GPT 5.4 Cyber, a more permissive model for defensive cybersecurity work, but only for vetted users.

The launch follows Anthropic’s restricted rollout of Claude Mythos Preview, a model aimed at finding serious software vulnerabilities and improving defenses.

Both companies are pushing more capable security-focused AI while tightly controlling access to reduce misuse in offensive cyberattacks or exploit development.