Sunday: when governments become the gatekeepers
The week's defining thread isn't a benchmark or a model card — it's who gets to decide who gets access. Two frontier labs handed that question to Washington, and OpenAI quietly previewed what comes next anyway.
U.S. government allows Anthropic to release restricted 'Mythos' AI model to trusted U.S. organizations
A government-mediated access tier for a frontier model is a new kind of infrastructure — one built from national security logic, not commercial terms.
U.S. government to vet and approve access to OpenAI's GPT-5.6
Government vetting of who can use a commercial model shifts the access question from terms-of-service to policy — with real consequences for international competitiveness.
OpenAI previews GPT-5.6 Sol, a next-generation model
A next-generation model preview, arriving just as access to the current one is being government-gated, shows how fast the frontier keeps moving regardless.
OpenAI unveils first custom AI chip, manufactured by Broadcom
Building custom silicon with Broadcom is a long-term bet on infrastructure independence — the kind of commitment that reshapes a lab's cost structure and strategic leverage.
Get the next edition in your inbox
Midweek and Sunday. Primary-sourced AI news paired with the guides that explain it.