In a fortnight, two leading AI companies, Anthropic and OpenAI, postponed the launch of their latest frontier AI models—Fable 5 and Mythos 5 for Anthropic, and GPT 5.6 for OpenAI—after the US government intervened. Anthropic had already launched Fable 5 and Mythos 5 for paying Claude customers worldwide before access was revoked on June 12th, as the US Department of Commerce restricted export to worldwide users, deeming the models' cyber capabilities a threat in the hands of potential adversaries despite guardrails. On June 25th, the White House asked OpenAI to delay its GPT-5.6 launch in stages, resulting in the latest ChatGPT model debuting only to a handful of government-vetted partners, not paying customers worldwide. OpenAI CEO Sam Altman told staff they were “approving access customer by customer” with US government help. Pro-safety group Public First’s Brad Carson called this a form of gatekeeping of cutting-edge capability not grounded in formalised laws or regulation, with justification possibly illegal. The US government points to Pax Silica, a US-led initiative from December 2025 to secure semiconductor chips and frontier AI models. AI pioneers Geoffrey Hinton and Yoshua Bengio, along with 22 co-authors, have warned against reckless AI advancements, though Bengio also warned that licensing and distribution restrictions concentrating power in a handful of labs threatens democracy and financial markets.