U.S. Government Intervention in GPT-5.6 Release Could Replay Fable Delisting Tragedy

According to Dynamic Beating monitoring, due to federal security review requirements, OpenAI CEO Sam Altman has confirmed a delay in the full release of GPT-5.6, opting instead to provide a limited preview to a few select partners. During the preview period, access will be granted to customers on a case-by-case basis with approval from the U.S. government. U.S. Secretary of Commerce Howard Lutnick also issued a warning to OpenAI, stating that a full release is not permitted until receiving interdepartmental joint approval. Altman has expressed to the government that a phased preview is not the preferred long-term model, and OpenAI will seek a more sustainable future release mechanism.

The U.S. government's intervention in cutting-edge large models has raised concerns in the industry about an implicit admission licensing regime. Earlier this month, Amazon CEO Andy Jassy reported a security vulnerability to the White House, leading to restrictions on overseas access to Anthropic's flagship models, prompting Anthropic to permanently take down Fable and Mythos.