Artificial Intelligence is here whether you like it or not. The question is no longer whether artificial intelligence will shape American life, but whether we will construct the legal architecture surrounding it to promote prosperity and productivity for all Americans—or permit it to become an instrument of nationwide persecution.
A patchwork of state AI laws and regulations is already attempting to ensure the latter. Because AI models are developed centrally and deployed uniformly across jurisdictions, rational actors will conform to the most stringent requirements anywhere on the books. The result is governance by the lowest common denominator: one radical legislature in California, New York, or Washington state could effectively dictate the operational parameters of every AI system used from coast to coast. Red state lawmakers and their constituents will have no meaningful voice, despite having vastly superior regulatory schemes that benefit the citizens of their states. This is not federalism; it is de facto national policy set by the most extreme blue-state majorities.
That is why President Trump’s call for federal preemption of conflicting state AI laws is not merely prudent—it is imperative. Absent a single, uniform national standard, the United States will not realize merit-based, truth-seeking AI that advances the common good. Instead, we will inherit the “woke AI” that the radical left is now embedding into disparate state regimes.
We already have an example of this happening.
Lawyers and business owners are intimately familiar with disparate-impact liability—a doctrine that, for decades, enabled regulators to sanction facially neutral, merit-based practices solely because their statistical outcomes failed to satisfy activist preferences. Last year, President Trump directed the federal government to eliminate the use of disparate-impact liability to the maximum extent permissible under law. In December, the Department of Justice promulgated regulations effectuating that directive, restoring the principle of equal treatment under neutral standards for every American.
Yet even as those federal safeguards are being implemented, blue states and municipalities are racing to codify the opposite regime into AI governance.
New York City’s Local Law 144, for instance, mandates annual bias audits and public disclosure of impact ratios by race and sex for any automated employment decision tool. Colorado’s SB 24-205 imposes civil penalties of tens of thousands of dollars for any covered entity that fails to exercise “reasonable care” to avoid “algorithmic discrimination” in employment, housing, credit, and other high-stakes domains. New Jersey’s 2025 regulations go further still: they expressly impose disparate-impact liability on employers utilizing AI tools. Where a facially neutral algorithm produces a statistically disparate outcome, the burden of proof shifts to the employer to justify its methodology—or face fines of thousands of dollars per violation. Critically, this strict liability attaches even where a small business merely deploys an off-the-shelf system it did not itself design or train.
The pattern is unmistakable. Having lost the policy debate over equal treatment, the left is attempting to smuggle disparate-impact theory through the back door of AI regulation. The consequences are predictable and pernicious: an AI-driven resume screener compelled to disregard criminal records, aptitude test results, or employment history whenever the data yield “wrong” demographic ratios. A military veteran from a Christian household may be silently filtered from consideration because the model was engineered to evade disparate-impact scrutiny. Merit will be subordinated to mandated outcomes. Productivity will yield to persecution.
This legal architecture will not be limited to hiring. Credit underwriting, mortgage origination, insurance pricing—every domain in which AI informs consequential decisions will be distorted by the same burden-shifting regime. Domestic innovators will expend scarce resources on perpetual compliance rather than genuine advancement. And because every model must satisfy the most aggressive jurisdiction, the entire nation will effectively operate under California’s rules, notwithstanding that most Americans never consented to them at the ballot box.
Some conservatives have expressed concern that federal preemption infringes upon traditional principles of states’ rights. Respectfully, that objection misapprehends the nature of the technology at issue. AI is quintessentially interstate: a model cannot be practically programmed to behave differently in New Jersey than in Texas without destroying its utility and scalability.
President Trump understands the stakes. His administration has already begun the work of extirpating radical legal theories from the federal bureaucracy. It must now complete the task by establishing a coherent national framework—one that expressly rejects disparate-impact liability in the AI context, safeguards merit-based decision-making, and ensures these transformative tools serve the American people rather than punish them.
The radical left is laboring assiduously to convert AI into the most sophisticated mechanism of social engineering yet devised. We can foreclose that outcome, but only if we decline to allow a handful of blue state legislatures to set the rules for the republic.
A single federal standard, grounded in the equal application of neutral merit-based criteria, is the sole means of guaranteeing that AI delivers prosperity and productivity, not persecution.
Gene Hamilton is president of America First Legal and a former deputy White House counsel. He previously served in senior positions at the Departments of Justice and Homeland Security.
