Wednesday brought a notable policy shift from Washington, a major model reveal from Microsoft, and fresh numbers underlining just how much capital is flowing into AI infrastructure — a reminder that the race is being run on regulatory, technical, and financial tracks simultaneously.
White House Issues Voluntary AI Oversight Executive Order
President Trump signed an executive order titled Promoting Advanced Artificial Intelligence Innovation and Security on June 2, asking AI companies to voluntarily submit frontier models for government review up to 30 days before public release. The order also directs agencies to build an “AI cybersecurity clearinghouse” and develop benchmarks for assessing powerful models’ cyber capabilities. Crucially, it explicitly bars any mandatory licensing or pre-clearance regime — keeping the framework opt-in. Read the full order and NPR’s analysis.
Microsoft Unveils MAI-Code-1-Flash and MAI-Thinking-1 at Build 2026
At its Build developer conference in San Francisco, Microsoft announced its first in-house frontier models: MAI-Code-1-Flash, a code-generation model that powers GitHub Copilot, and MAI-Thinking-1, a high-efficiency reasoning model targeting low token costs. The company also previewed updated cloud models for speech, voice synthesis, and image generation, plus small Aion models designed to run locally on Windows PCs. The move signals Microsoft’s intent to reduce its reliance on OpenAI while expanding its AI product surface. Details from CNBC.
$36 Billion Private Credit Deal Fuels Anthropic’s Compute Expansion
Apollo Global Management and Blackstone are structuring a $36 billion private credit deal to purchase Google custom TPU chips on Anthropic’s behalf, routed through a special-purpose vehicle that leases the chips to Anthropic’s data centers. The deal stacks on top of Google’s separate up-to-$40 billion equity investment (announced in April at a $350 billion valuation) and a parallel 5 GW compute commitment from AWS — giving Anthropic a combined 10 GW footprint. Anthropic’s annualized revenue has surpassed $30 billion, up from roughly $9 billion at end-2025. TechCrunch on the Google deal.
GPT-5.6 Spotted in OpenAI’s Codex Logs
References to a GPT-5.6 model surfaced in OpenAI’s Codex tooling logs, fueling speculation about an imminent release. Prediction markets are pricing the probability of a June 30 launch above 80%. GPT-5.5 (released in April) and its lighter variant GPT-5.5 Instant (released in May) are already the default models in ChatGPT, so the next jump would mark OpenAI’s third major model update in under three months. No official announcement has been made yet. WaveSpeed on the Codex leak.
What to watch: How many AI labs actually opt into the White House’s voluntary review framework will be the real test of whether the executive order shapes frontier model development — or remains largely symbolic.