Technology
Anthropic, OpenAI and Google deepen their federal-frontier-model gating regime as analysts watch open weights
Following the limited customer-by-customer release of OpenAI's GPT-5.6 to US-vetted partners and Anthropic's clearance of Claude Mythos 5 to roughly 100 trusted US organisations, analysts are now framing the two-tier release pattern as a de facto US frontier-model gate. The conversation has expanded to whether open-weight releases above a defined capability line will face the same preclearance regime, with a Stack Futures tracker counting two labs gated within a single seven-day window under voluntary White House review.