Blackburn's TRUMP AMERICA AI Act would erase all state AI laws with one federal bill

Senator Marsha Blackburn (R-TN) released a section-by-section summary on December 19, 2025 of the TRUMP AMERICA AI Act. The acronym stands for The Republic Unifying Meritocratic Performance Advancing Machine Intelligence by Eliminating Regulatory Interstate Chaos Across American Industry Act—a rhetorical device common in Congress where bill names are crafted to create memorable abbreviations that signal political alignment. As of February 2026, only the summary exists; the full bill text has not been formally introduced.

Blackburn organizes the bill around a "4 Cs" framework

Children: incorporates the Kids Online Safety Act (KOSA) provisions protecting minors online

Creators: includes the NO FAKES Act, requiring consent before AI systems train on creative works Conservatives: adds political affiliation as a protected class in AI bias evaluations, treating political viewpoint discrimination the same as racial or gender discrimination for AI purposes Communities: establishes a duty of care requiring AI developers to take reasonable precautions against foreseeable harms.

Adding political affiliation as a protected class in AI bias evaluations is unprecedented in federal civil rights law. No existing federal statute includes political viewpoint as a protected class. Critics argue this provision could weaponize anti-discrimination law to prevent platforms from moderating political misinformation, since treating all political viewpoints as equally protected could constrain platform content moderation decisions.

The "Bad Samaritan" provision rewrites Section 230 of the Communications Decency Act, the 1996 law that lets internet platforms host user-generated content without being liable for it. Section 230 currently gives platforms broad immunity for content moderation decisions. Blackburn's provision would strip immunity from platforms that "purposefully facilitate or solicit" content violating federal criminal law—shifting the litigation posture by requiring platforms to prove in court that they did not facilitate illegal content.

Federal preemption of all state AI laws is the bill's most sweeping provision

If enacted, it would override 🏔️Colorado's AI Act, 🌽Illinois's HB 3773 employment law, and every other state-level AI regulation

This would eliminate the "laboratory of democracy" approach where states experiment with different regulatory models Tech industry leaders, who prefer uniform federal rules over a patchwork of 50 state requirements, benefit most from preemption; state consumer protection advocates lose leverage.

Data centers exceeding 100 megawatts would bypass standard environmental impact review under a fast-tracking provision. Facilities of this scale consume electricity equivalent to roughly 80,000 homes and require massive water resources for cooling. Environmental groups argue that bypassing environmental review could harm local communities through water depletion, carbon emissions, and habitat disruption—but AI companies framing data centers as national infrastructure compete directly with those environmental concerns.

Codifying Executive Order 14365—which Trump signed December 11, 2025—into federal statute would make its AI development directives permanent and resistant to reversal by future administrations. Executive orders can be rescinded by subsequent presidents with a stroke of a pen, but statutory codification requires an Act of Congress to undo. This converts temporary executive policy into durable law.

The legislation's bundling strategy—combining KOSA, the NO FAKES Act, Section 230 reform, federal preemption, and data center fast-tracking into a single bill—creates complex political dynamics. Supporters of children's online safety who oppose federal AI preemption must vote for or against the whole package. This all-or-nothing structure forces legislators and stakeholders to negotiate across policy areas they'd normally address separately.