White House science adviser Michael Kratsios unveils first comprehensive congressional AI framework, targeting national standard to replace state patchwork

Summary

Michael Kratsios on the White House AI Legislative Framework

Michael Kratsios, Director of the White House Office of Science and Technology Policy, has unveiled what he describes as the first comprehensive AI legislative framework ever presented by the White House to Congress. The central goal is a single national AI regulatory standard that would preempt the growing patchwork of state-level laws.

The core problem it addresses

Kratsios argues that large companies can navigate conflicting state AI regulations, but small startups cannot. A federal standard would override state-level rules in covered areas, blocking measures like New York's proposed ban on AI legal advice, which Kratsios says the White House is "very against." His framing is straightforward: federal preemption protects the ability of Americans to access AI tools, including ones that give ordinary people access to legal or professional guidance they couldn't otherwise afford.

What the framework covers

Beyond preemption, the framework spans five areas:

• Child safety — Congress should require platforms to give parents tools to see and control what their children experience online. Kratsios calls this an "80-20 issue" with bipartisan support and expects it to be a legislative on-ramp for the broader framework.
• Ratepayer protection — The White House is urging Congress to codify the ratepayer protection pledge, under which major AI companies committed to build, bring, or buy their own power for any data center they construct. The Biden administration had banned behind-the-meter power generation; the Trump administration reversed that via executive action and wants it enshrined in law.
• Worker upskilling — Existing Labor Department, Education Department, and Small Business Administration programs should be explicitly reoriented toward preparing workers for an AI economy.
• AI in education — Referenced but not detailed in the conversation.
• National security evaluation — Rather than creating a regulatory body that can block or nationalize AI companies based on safety tests, the framework calls for relevant national security agencies to build independent expertise to evaluate frontier models from a defense and cybersecurity standpoint.

Timeline

Kratsios says the White House hopes to get this through Congress within calendar year 2026. Child safety provisions, given their bipartisan history, are the likeliest vehicle to move first and pull the rest of the framework along.

The small business argument

Kratsios pushes back on the idea that AI's economic impact is primarily a large-enterprise story. The vast majority of Americans work at small businesses, he argues, and the productivity gains from accessible AI tools are most transformative at the five-to-50-person scale — a business that can now handle inbound calls, qualify clients, and schedule jobs without hiring additional staff.

The regulatory philosophy shift

The clearest line Kratsios draws is between the current administration's approach and the prior one. The Biden-era framing, in his telling, placed AI evaluation inside regulatory bodies with the authority to restrict or nationalize companies that failed safety tests. The Trump administration's framework limits government's role to national security assessment — letting agencies supporting warfighters and cybersecurity evaluate models, without granting those agencies commercial veto power.

Whether Congress moves the full framework or just the child safety provisions will determine how much of this philosophy actually becomes law before the end of the year.