OpenAI Builds Exclusive Cybersecurity AI with 'Trusted Access'

OpenAI is reportedly developing a powerful new cybersecurity AI product, restricting access to a select group of defensive security operators through its 'Trusted Access for Cyber' program. This move mirrors Anthropic's approach with its 'Claude Mythos' model, signaling a shift towards controlled distribution of advanced AI.

2026-04-10T01:56:52.784Z

OpenAI is developing a new cybersecurity product to be exclusively released through its "Trusted Access for Cyber" program. This initiative aims to keep highly capable AI tools out of public reach, providing them only to defensive security operators.

The program follows the release of GPT-5.3-Codex, OpenAI's most advanced cybersecurity offering to date, with participants receiving $10 million in API credits. This development comes amid widespread concern over advanced AI's potential to exploit system vulnerabilities.

Earlier, Anthropic restricted access to its "Claude Mythos" model, deeming it too effective at finding zero-day vulnerabilities across major operating systems and browsers. This restricted access strategy is seen as a response to increasing regulatory scrutiny and a proactive measure to position companies as responsible actors in the AI space.

The trend suggests that frontier AI models will increasingly be distributed as select, controlled releases rather than broad public launches, prioritizing responsible use by vetted organizations.