OpenAI Restricts Access to Cyber

OpenAI will roll out its security-focused model GPT-5.5 Cyber to critical cyber defenders on a priority basis within a few days.

Altman announced this on X, saying that access will only be reviewed for users who submit their credentials and intended use plan through an application form on the OpenAI website.

Cyber is designed to help with the following tasks.

• Penetration testing
• Vulnerability identification and exploitation
• Malware reverse engineering

The key point is that OpenAI adopted the same kind of access control right after Altman criticized Anthropic for limiting Mythos to only select users.

OpenAI is consulting with the US government and continuing to look for ways to make it more broadly available to users with formal cybersecurity credentials.