Washington becomes the gatekeeper for America's top AI models

Sunday, June 28, 2026

The biggest AI story today isn't a model release — it's who controls access to one. The US government has quietly made itself the arbiter of who can run the most capable systems from both Anthropic and OpenAI.

Trump Administration Becomes Gatekeeper for the Most Powerful New AI Models

Both OpenAI and Anthropic agreed to restrict their most advanced models at the Trump administration's request following a cybersecurity review, making Washington a de facto gatekeeper for frontier AI. It's a precedent-setting shift: the most powerful models are now governed less by their makers than by the government's national-security calculus.

Read the brief →

US Clears Anthropic to Switch Claude Mythos 5 Back On — for a Vetted Few

The US cleared Anthropic to bring its Claude Mythos 5 model back online after more than two weeks dark, but only for a tightly vetted set of organizations rather than the public. The selective restoration shows the new model of access — frontier capability is becoming a controlled good, doled out to approved users.

Read the brief →

White House Pushes OpenAI to Lock GPT 5.6 to Vetted Government Partners

The White House is pushing OpenAI to limit its newly previewed GPT-5.6 to vetted government partners, mirroring the constraints already placed on Anthropic. Read together with the Anthropic move, it signals a coordinated policy of gating the strongest models behind federal approval.

Read the brief →

OpenAI Unveils GPT-5.6 Series: Meet Sol, Terra, and Luna

OpenAI launched a limited preview of its GPT-5.6 series — a three-tier lineup of Sol (flagship), Terra (balanced), and Luna — the very models now caught up in the access fight. The release matters because the frontier capability jump is exactly what's drawing government scrutiny.

Read the brief →

China's AI Challengers Claim to Match Anthropic's Banned Models

Chinese firms are pitching their latest systems as direct rivals to the American models now walled off by export controls and federal gating. If the claims hold, US restrictions may accelerate exactly the homegrown Chinese competition they were meant to contain.

Read the brief →

AI Spending Booms, But Analysts Flag a Valuation Hangover Ahead

Enterprise AI spending is surging in 2026 on the back of agent-driven automation, but Morgan Stanley and other analysts are flagging a looming valuation hangover. The split — booming demand, stretched prices — is the tension that's been whipsawing tech stocks all week.

Read the brief →

Apple's Vision Pro Chief Paul Meade Defects to OpenAI's Hardware Team

Paul Meade, Apple's VP overseeing Vision Pro and its smart-glasses effort, is leaving to join OpenAI's hardware team. It's a marquee defection that underscores OpenAI's serious ambitions in consumer devices — and Apple's struggle to hold elite hardware talent.

Read the brief →

Alphabet's AI Brain Drain Spooks Investors Even as It Joins the Dow

A wave of departing AI researchers is rattling Alphabet investors even as the company joins the Dow, fueling fears that Google is losing the talent race at the frontier. Brain drain at this level is a leading indicator of where the next breakthroughs — and valuations — will land.

Read the brief →

Apple Reportedly Lobbies Washington to Buy Cheaper Chinese Memory Chips

Apple is quietly lobbying Washington for permission to buy memory chips from a Chinese maker flagged for military ties, per the FT. The request highlights the awkward collision between supply-chain economics and the same national-security politics now reshaping AI access.

Read the brief →

The AI Briefing The day's most important AI, chip & tech stories — one email each morning. Free.