The Trump administration has reportedly told OpenAI to hold back the release of its latest model, GPT 5.6, and distribute it only to a select group of partners for now.

OpenAI's CEO Sam Altman is said to have informed staff that access would be granted on an individual basis during a preview period, with hopes of a wider release following if all goes well. This move mirrors Anthropic’s earlier decision to limit distribution of its powerful AI model, Claude Mythos.

The concern stems from the potential misuse of such advanced tools by cybercriminals. AI models like Cyberfrontier can write malware and execute ransomware attacks autonomously. They are also capable of rapidly identifying and exploiting software vulnerabilities at a pace that surpasses human analysts.

While Anthropic’s approach was met with controversy, this move from OpenAI suggests the government is taking a more proactive role in regulating AI deployment. This raises questions about the balance between innovation and security as AI continues to evolve.