In the Wake of Anthropic’s Mythos, OpenAI Has a New Cybersecurity Model—and Strategy

OpenAI unveiled GPT-5.4-Cyber, a new cybersecurity-focused model aimed at helping digital defenders.

The company paired the launch with a three-part strategy: customer validation, staged releases, and broader security investments.

OpenAI’s approach emphasizes trusted access and iterative deployment to limit misuse while still giving defenders access to advanced capabilities.

The move contrasts with Anthropic’s more restrictive rollout of Claude Mythos Preview, highlighting different AI safety and release philosophies.