GPT 5.6 Limited Release: White House Cites Safety Fears

The White House has asked OpenAI to restrict the launch of GPT 5.6, meaning it will only be available to individually approved partners initially. This marks the first time the government has directly influenced the rollout schedule of a new AI model. GPT 5.6 is seen as having capabilities similar to Anthropic's restricted Mythos model. Two White House offices, along with Commerce Secretary Howard Lutnick, were involved in this decision, citing cybersecurity concerns. OpenAI CEO Sam Altman shared this plan with employees, stating the government will approve access on a per-customer basis. He hopes for a broader public release within weeks if the preview period goes well, though he noted this is not the company's preferred long-term launch model. The bottom line is that enterprise customers now face regulatory risks when building around any single frontier AI model.