When autonomous driving startup PerceptIn got down to navigate America’s synthetic intelligence laws, it budgeted $10,000 for compliance. The precise invoice exceeded $344,000 per deployment challenge, greater than double the corporate’s analysis and improvement prices. Two months in the past, PerceptIn went out of enterprise.
Final yr, states launched greater than 1,200 AI-related payments, with at the least 145 changing into legislation, creating contradictory necessities that multiply compliance burdens. Every jurisdiction defines “artificial intelligence,” “high-risk systems,” and “consequential decisions” in another way, forcing firms to investigate equivalent know-how beneath a number of incompatible frameworks. A hiring device that satisfies California’s four-year recordkeeping and anti-bias testing necessities should additionally meet Colorado’s separate influence evaluation mandates and New York Metropolis’s unbiased bias audit regime with distinct discover necessities. Every jurisdiction defines core phrases in another way, forcing firms to investigate equivalent methods beneath a number of incompatible frameworks.
Trade estimates recommend compliance prices add roughly 17 % overhead to AI system bills. For small companies, California’s privateness and cybersecurity necessities alone impose almost $16,000 in annual compliance prices. However these figures understate the true burden as a result of they deal with compliance as a variable price that scales with firm measurement. The truth is way worse.
Harvard Kennedy Faculty researchers recognized what they termed a “compliance trap” by which regulatory prices devour sources quicker than startups can generate income. Their evaluation discovered {that a} 200 % enhance in mounted compliance prices transforms a startup’s working margin from constructive 13 % to unfavourable 7 %. That’s not a rounding error, it’s the distinction between survival and chapter. A 3-person group constructing an employment screening device faces the identical baseline compliance obligations as a thousand-person enterprise in lots of jurisdictions, however with out the income base or authorized infrastructure to soak up these prices.
This dynamic arms an unlimited aggressive benefit to the very firms state laws purport to constrain. Incumbent tech giants preserve compliance departments that dwarf whole startups. They will afford multi-jurisdictional authorized groups, customized bias auditing frameworks, and the political relationships essential to form rising necessities. For them, state AI fragmentation represents a manageable price of doing enterprise. For startups, nonetheless, it represents an insurmountable barrier to entry.
The strategic implications are staggering. Whereas American entrepreneurs waste engineering expertise on contradictory compliance regimes, Chinese language AI firms function beneath a unified nationwide framework. Beijing’s strategy is hardly a mannequin of light-touch regulation, but it surely provides what America’s patchwork can’t: coherent guidelines that don’t change at state traces. When compliance prices exceed improvement budgets, innovation doesn’t sluggish—it stops altogether or strikes to overseas jurisdictions the place the principles are clear.
The White Home acknowledged this aggressive hazard in a December 2025 government order criticizing the “patchwork of 50 different regulatory regimes” and directing the U.S. Division of Justice to determine an AI Litigation Activity Drive to problem state legal guidelines that hinder nationwide AI coverage. The order represents a needed first step, however government motion alone can’t resolve an issue rooted in legislative fragmentation. Congress should act.
Federal preemption laws would set up uniform nationwide requirements for AI methods whereas preserving states’ capacity to implement normal shopper safety legal guidelines. This isn’t a radical idea. Federal frameworks already govern aviation security, pharmaceutical approvals, and telecommunications—industries the place state-by-state variation would create equivalent chaos. Protected harbor provisions may defend firms that implement affordable bias testing and influence assessments from legal responsibility, creating incentives for accountable improvement with out imposing contradictory mandates.
The present trajectory is unsustainable. Each month that passes with this regulatory chaos intact represents one other month of American innovation surrendered to better-organized opponents. State legislators designed their AI frameworks to constrain Large Tech’s energy. As a substitute, they’ve constructed a moat round incumbents whereas crushing the startups that may problem them. The compliance entice isn’t defending shoppers. It’s defending monopolies and handing a aggressive benefit to overseas adversaries. That’s not regulation, it’s self-sabotage.
The opinions expressed in Fortune.com commentary items are solely the views of their authors and don’t essentially replicate the opinions and beliefs of Fortune.
