OpenAI will begin a public rollout of GPT-5.6 this week after the Department of Commerce cleared additional testing and consultations on national security risks.The model, offered in three variants—Sol, Terra and Luna—was previously limited to a small group of US partners at the White House’s request while Anthropic faced similar US-only curbs on Fable 5 and Mythos 5.Security researchers and lawmakers warn that
GPT-5.6’s capacity to identify software vulnerabilities could accelerate hacking and military misuse, while
OpenAI says added safeguards and oversight will accompany the expansion.