Trump Administration Restricts OpenAI GPT-5.6 Model Rollout

The Trump administration has requested that OpenAI restrict the initial launch of its GPT-5.6 model family, which includes the flagship Sol, mid-tier Terra, and high-speed Luna. Under this agreement, OpenAI is providing a limited preview to approximately 20 trusted partners, with federal officials approving access on a customer-by-customer basis. This intervention follows a June 2 executive order establishing a voluntary 30-day federal vetting period for powerful AI models to identify cybersecurity risks.

Commerce Secretary Howard Lutnick coordinated the effort after meetings with OpenAI CEO Sam Altman. The government's caution is driven by concerns that frontier models could be used by cybercriminals or nation-states to exploit software vulnerabilities. This follows similar pressure on Anthropic, which was forced to disable its Fable 5 and Mythos 5 models over national security threats and foreign national access.

OpenAI maintains that Sol is designed to fix vulnerabilities rather than execute attacks, and notes that the model did not cross its Cyber Critical threshold in testing. Despite cooperating to establish a repeatable process for future releases, the company cautioned that government-led access should not become a long-term default, as it hinders global availability for developers and enterprises. Rep. Lori Trahan criticized the move, arguing that the administration is deciding access without legal oversight or a clear process.