In the coming weeks, the United States Senate is expected to ramp up consideration of a sprawling budget bill passed by the House that, if adopted, could block states from enforcing artificial intelligence regulations for 10 years.

Hundreds of state lawmakers and advocacy groups have opposed the provision, which House Republicans approved last month as an attempt to do away with what they call a cumbersome patchwork of AI rules sprouting up nationwide that could bog down innovation. On Thursday, Senate lawmakers released a version of the bill that would keep the moratorium in place while linking the restrictions to federal broadband subsidies.

Critics have argued that the federal moratorium — despite carving out some state laws — would preempt a wide array of existing regulations, including rules around AI in healthcare, algorithmic discrimination, harmful deepfakes, and online child abuse. Still, legal experts have warned that there is significant uncertainty around which specific laws would be preempted by the bill.

To that end, one non-profit organization that opposes the moratorium on Friday is releasing new research examining which state AI laws would be most at risk if the moratorium is adopted, which the group shared in advance with Tech Policy Press.

The report by Americans for Responsible Innovation — a 501(c)(4) that has received funding from Open Philanthropy and the Omidyar Network, among others — rates the chances of over a dozen state laws being blocked by a moratorium, from “likely” to “possible” to “unlikely.”

Here’s a breakdown of what they found:

The laws likeliest to be blocked

The federal moratorium would “likely” sweep up a “wide range of public interest state AI legislation,” according to the report, including laws imposing transparency requirements on AI services or aimed at tackling algorithmic discrimination, including in the workplace.

That includes the Colorado AI Act, SB 24-205, a sweeping ,first-of-its-kind law set to usher in a new reasonable duty of care standard requiring those who develop or deploy high-risk AI systems to prevent algorithmic discrimination and take steps to mitigate against harms. The group said it would “likely be voided” because it “regulates AI technologies by way of establishing a civil liability scheme for developers of high-risk AI systems.”

A freeze on state AI enforcement would likely also catch laws in Utah and California aimed at boosting transparency around the tool’s use. That includes Utah’s Artificial Intelligence Policy Act, SB 149, which requires that it be disclosed to consumers when they are interacting with some generative AI products; and the California Artificial Intelligence Transparency Act, SB 942, which mandates clear disclosures for AI-generated audio and visual materials.

A pair of consumer protection laws in Illinois intended to address AI’s use in the workplace are also at high risk of being preempted, according to the report. HB 3773, which amends the state’s human rights law, prohibits using AI to make hiring decisions if it leads to discrimination, while the Video Interview Act, HB 2557, requires employers to notify applicants when they deploy AI and obtain their consent before using it for evaluations.

The laws that could be blocked

For some state laws dealing with AI, how the moratorium would apply is less clear-cut. According to the report, its “plain language could also potentially capture laws addressing social media systems, data privacy protections, and other algorithmic-based technologies.”

That could “possibly” include data privacy laws in Texas, Maine, and Connecticut. The Texas Data Privacy and Security Act, HB 4, for one, “appears to regulate covered AI technologies by way of creating consumer rights to opt out of profiling," and thus could be voided. Maine’s LD 1585, meanwhile, would likely target facial recognition technology and contains restrictions on third-party agreements that “would potentially constitute regulating AI technologies.” And Connecticut’s SB 3 could be nullified because it “restricts certain social media system design features and automated targeted advertising,” Americans for Responsible Innovation found.

Child online safety initiatives are poised to run into a similar limbo. New York’s landmark attempt to ban “addictive” social media algorithms, the Stop Addictive Feeds Exploitation For Kids Act, SB 7694A, could be blocked because of its definition of algorithmic feeds, the group wrote. So too could the Utah Minor Protection in Social Media Act, SB 194, which sets restrictions on certain platform features that can lead to excessive use.

The group said another bucket of laws that may be impacted are ones dealing with “generative AI harms,” such as Tennessee's ELVIS Act, SB 2096, which prohibits the use of AI to create an unauthorized deepfake of another individual, and Virginia’s Synthetic Digital Content Act, HB 2124, which criminalized the use of synthetic media to defraud or defame someone.

The laws likeliest to dodge the moratorium

According to the report, the laws that are least likely to be obstructed largely deal with how the technology is incorporated by state governments. That’s because during the negotiations in the House, lawmakers inserted language stipulating that only regulations dealing with AI that are “entered into interstate commerce” would be affected, which in-house state work likely is not.

Because of that, measures like California’s Generative Artificial Intelligence Accountability Act, SB 896, and New Hampshire’s HB 1688, which both set out rules for how state agencies and departments ought to deploy the technology, are “unlikely” to be affected, the report found.

Status in limbo

The report was based on the moratorium provision as it was written into the reconciliation package that passed the House last month, but it remains to be seen if the AI provision will survive in the Senate, including a review of whether it will be eligible under the “Byrd Rule.”

The Senate Commerce Committee on Thursday released its version of the reconciliation package, which contained a revised 10-year moratorium on state AI laws. Under this iteration, states would be blocked from receiving federal broadband subsidies if they do not temporarily halt their enforcement of state AI laws during the period. A "talking points" memo released by Senate Commerce Chairman Ted Cruz (R-TX) said the bill would block states from collecting broadband funds that were "strangling AI deployment with EU-style regulation."

By tying the provision more concretely to the disbursement of federal funds, the revised version could be more likely to withstand the Byrd Rule, which states that legislation take up through reconciliation cannot be "extraneous" to the budget itself.

It’s also unclear whether it will have enough political support in the chamber, where Democrats and some Republicans, such as Sen. Marsha Blackburn (R-TN), have already pushed back on the idea. Republicans are hoping to finalize the reconciliation package before the Fourth of July.