US Government's Pre-Release Authority Over Frontier AI Becomes Operational; OpenAI's Equity-Stake Proposal Advances in Parallel

The Trump administration’s June 2 executive order requiring up to 30 days of pre-release government access to frontier AI models is now operational: OpenAI delayed the full public launch of GPT-5.6 after the government invoked the authority, and Anthropic’s Fable 5 and Mythos 5 were taken offline under the June export-control directive before being restored first to critical-infrastructure operators. Anthropic says it will now coordinate with government on all future frontier releases. These moves mark a shift from voluntary pre-deployment safety commitments to binding government oversight of AI capability deployment, running concurrently with OpenAI’s proposed 5% government equity stake (~$42.6B) reported the same week (2026-07-02–openai-proposes-5-percent-government-equity-stake).

Together, binding pre-release control and a proposed equity stake describe two simultaneous vectors of state entanglement with frontier AI labs — operational control over what ships, and a financial stake in the companies shipping it — extending the export-control precedent set against Anthropic in June (2026-06-12–commerce-export-control-order-forces-anthropic-to-suspend-fable-5-mythos-5-worldwide) from a punitive one-off action into a routine, institutionalized government checkpoint on frontier model releases.

Update (2026-07-28): OpenAI and Anthropic are co-designing, alongside NIST’s Center for AI Standards and Innovation (CAISI, the renamed AI Safety Institute), the capability-threshold definitions that determine which frontier models face this pre-release scrutiny. CAISI’s pre-deployment testing agreements have since expanded to Google DeepMind, Microsoft, and xAI as of May 5, but OpenAI and Anthropic remain the two labs embedded in defining the thresholds themselves — a structural advantage over rivals who were not in the room when the rules were written. This is regulatory capture through participation rather than lobbying: the same labs whose models are subject to the checkpoint helped author what triggers it. Independently confirmed by TechTimes and NIST’s own public agreements documentation.