The AI Moratorium Could Gut State Tax Revenues and Give Fraudsters a Pass

David Brody is a civil rights and technology legal expert affiliated with the Multiracial Democracy Project at George Washington University Law School.

Update from author: On the morning of June 25, after the publication of this piece, the office of Senator Ted Cruz (R-TX) released revised text for the AI moratorium, adding the word "or" between the two exception clauses in the provision. While the exact implications of these changes require further review, they appear to allow states to enforce generally applicable laws—such as tax or criminal laws—that apply broadly to all sectors, including AI and other tech systems. However, states would be prohibited from enforcing laws that specifically target the tech industry. This means state-level taxes on data centers, rideshare apps, or online commerce would still not be permitted, as would laws specifically aimed at online crimes or robocall scams committed through use of a computer.

Tucked away in an innocuous-sounding provision of the Big Beautiful Bill is a landmine that could blow up state tax revenues while also giving cybercriminals a get-out-of-jail-free card. The bill contains a provision that blocks states from enforcing any law that regulates tech if they want to access federal funding to support broadband deployment. While the provision is styled as a “pause” on AI regulation, that is a Trojan horse. The definitions and other text are written in a manner that would apply to almost any form of computing, not just AI. In addition, the listed exceptions are so narrowly scoped and poorly drafted that they apply to almost no state policy.

The provision is also written so broadly that it applies to all types of state laws, including state tax law. The so-called AI moratorium could have devastating effects on state and local tax revenues, which would likely dwarf the funding states receive for broadband. It could block state tax laws on online commerce, software-as-a-service businesses, sales of tech products and services, rideshare apps, food and grocery delivery apps, data center property taxes, sector-specific excise taxes, and other activities where the sale or use of tech is a significant component of the economic activity. This preemption of state and local tax laws will incentivize gamesmanship by big corporations and disproportionately harm small businesses that cannot afford to add expensive software suites to their workflow just to avoid paying taxes.

The exceptions in the moratorium will not insulate state and local tax law. To fall within the exception clauses, a state law must both be directed at reducing burdens on tech and be a tech-neutral, generally applicable provision. That second requirement explicitly calls out tech-sector “taxation” as impermissible. Even if there may be ambiguity in how many state tax laws get preempted, states and localities will be tied up in years of litigation to sort this out and collect the tax revenues they are due. Many companies could exploit this new federal loophole to evade taxes; state and local governments may not have the resources to chase them all.

The AI moratorium also blocks states from prosecuting fraudsters and scammers targeting seniors, families, veterans, children, and anyone else. Unlike the version that passed the House of Representatives, the Senate version of the moratorium does not have a standalone exception for state criminal laws. States are the primary enforcers of criminal law; only a small minority of crimes are prosecuted by the federal government. This includes prosecutions for fraud, theft, and scams—states lead the way. This moratorium will block the enforcement of state criminal law whenever the crime is committed through a computer.

That means criminals engaged in online scams, robocall scams, bank fraud, identity theft, and many other forms of theft will have a get-out-of-jail-free card to block state prosecutors. Even if the definitions in the moratorium are read narrowly to only apply to advanced forms of AI, such as ChatGPT, those systems are still exempt from criminal law. AI-enabled deepfakes and impersonations are already a serious fraud threat for consumers, banks, and other businesses. Making those fraudsters above the law will pour gasoline on the fire.

The moratorium also creates perverse incentives for bad actors to deliberately add technology to their workflow in order to avoid accountability. Want to rip someone off? Add Anthropic’s Claude AI or another software suite to your workflow, and now any state enforcement action against you is forbidden.

Now, some have argued that the moratorium only applies to a small pot of $500 million in new funding for states to deploy AI. This is not true. It applies to the entire amount of remaining funds from the $43 billion allocated to the Broadband, Equity, Access, and Deployment (“BEAD”) program, the vast majority of which has not been disbursed to the states.

The revised text of the provision says it applies to any BEAD funds granted after the date of enactment. But states have not yet received final approval from the Commerce Department on their broadband proposals, meaning the funds have not been released to them, and states have yet to move forward with contracts to subgrantees to initiate broadband projects. Moreover, President Trump’s Commerce Secretary Howard Lutnick released new guidelines for BEAD applications earlier this month, compelling the states to resubmit their applications to the agency. Should the Senate’s AI moratorium pass, states would then also have to certify compliance with the moratorium before Commerce releases the funds.

This moratorium puts states in an impossible position: forgo billions of dollars in broadband funding for rural and low-income consumers or make Big Tech above the law. The former exacerbates the digital divide. The latter decimates state tax revenues, allows cybercriminals to run roughshod over seniors and families, and guts state efforts to protect civil rights, consumers, online safety, and privacy.