What is GPT-5.4-Cyber and who can use it?

Vetted security pros, vendors, researchers via OpenAI's tiered Trusted Access program—starts narrow, scales up.

How does OpenAI prevent cyber AI misuse?

Strict identity verification, query monitoring, and phased rollouts; focuses on users over model limits.

Will OpenAI give GPT-5.4-Cyber to the US government?

Not yet—discussions ongoing with internal safety reviews.

OpenAI's GPT-5.4-Cyber: Wider Access Amid Hacking Fears [2026]

OpenAI's flipping the script on cyber AI: more power to vetted defenders, less blanket restrictions. But as hacking risks explode, is this trust-and-verify gamble genius or reckless?

Supply Chain Beat Apr 15, 2026 4 min read

⚡ Key Takeaways

OpenAI shifts from restricting AI capabilities to verifying users, expanding GPT-5.4-Cyber access to thousands. 𝕏
Unlike Anthropic's tight Mythos lockdown, OpenAI prioritizes defensive cyber tools with tiered permissions. 𝕏
Compute demands and verification delays could limit broad adoption despite the hype. 𝕏