LOWELL — UMass Lowell College Democrats President Matthew Kelly appreciates that U.S. Rep. Lori Trahan filed a bill to do something about regulating the growing artificial intelligence industry, but to him and a small group of protesters at Trahan’s office Thursday, part of the Frontier Act could have negative consequences.

Trahan, a Democrat, and a group of other bipartisan lawmakers introduced the Frontier Act last month with the stated goal of creating a “risk-based framework” for regulating the young but rapidly growing AI industry. Trahan doubled down on her support for the bill earlier this month after multiple major AI firms announced incidents in which their AI models were able to hack into other firms on their own after escaping from their testing environments.

The Frontier Act, Trahan said earlier this month while touting the bill, would establish “tiered requirements based on the size of a frontier AI developer, including model cards, risk-management frameworks, independent audits, incident reporting, and ongoing assessments.”

While Kelly and the other protesters were supportive of AI regulation, they took issue with a section within it they call “preemption,” which they interpret as being a prohibition of individual states and municipalities creating their own AI regulations beyond those imposed by the federal government. When the group arrived at Trahan’s office, they did so with a collection of more than 4,000 signatures on a petition opposing the Frontier Act because of the concerns over preemption. The demonstration was a partnership between organizations like QuitGPT, the College Democrats of Massachusetts, Third Act Massachusetts, Young People’s Alliance of Massachusetts, Voters of Tomorrow Boston and Massachusetts 50501.

“Including preemption, banning states from implementing any law in the future about regulating AI and what works best for their constituents is not progress,” said Kelly, calling it instead a “sellout” to big AI firms and the Trump administration.

“This bill is not going to help anybody, it is only going to allow for the furthering of the corporate interest, and at the end of the day this is another example of how politicians are selling out to the Markley corporation, that are trying to impede on the civilians of our city and building out data centers that impact everybody in a negative way.”

The section of the bill in question reads “PREEMPTION.—Except as provided in subsection (c), no State or political subdivision of a State may adopt or enforce any law, regulation, order, or other requirement that imposes new substantive obligations on artificial intelligence developers with respect to any Covered Subject Area.”

Immediately following that clause is another section which lays out what powers states or municipalities do have under this bill. That includes the ability to “adopt or enforce generally applicable laws, regulations, orders, or other requirements that do not target artificial intelligence developers,” and to “regulate the use or deployment of AI systems by deployers or users, including via consumer protection, civil rights, contract, criminal, or privacy laws, provided that no substantive obligations are imposed on developers with respect to model development, training, evaluation, or release.”

It also explicitly allows states and municipalities to create laws and regulations around the protection of minors from the possible harms of AI, and the use of AI by state governments.

In response to a request for comment from The Sun, Trahan’s office denied that the bill would ban states from passing AI laws. In a newsletter she sent to constituents Wednesday, Trahan addressed this very issue, and called the protesters’ interpretation “a misunderstanding of what the bill actually does.”

“The FRONTIER Act does not ban states from regulating artificial intelligence. In fact, it expressly preserves state authority to protect kids online, and to enforce the civil rights, consumer protection, and privacy laws that govern how AI is developed, deployed, and used,” said Trahan. “What it does is create a narrow federal lane for the world’s most powerful models – mandatory incident reporting, independent verification, and an emergency backstop to prevent catastrophic risk. States keep their broader authority to protect the people who live there.”

Trahan said she believes some of the confusion stems from “lumping our targeted bill in with the sweeping, industry-backed proposals to freeze state AI laws across the board.”

“Those are not the same thing, and I don’t support wiping out the protections states have already put in place,” said Trahan.

She referenced recent hacking incidents by AI models being tested by OpenAI and Anthropic, and said the Frontier Act would make it mandatory, not voluntary, for the companies to report such incidents within 72 hours.

In a recent Q&A session about the same topic on Instagram earlier this week, Trahan laid out areas of the bill where state and local governments would be preempted from making new laws, but where the bill also sets new regulations and standards on its own. That includes independent auditing where outside experts would be required to inspect AI models and report findings, incident reporting standards, mandatory risk-management plans and emergency stop orders that could force AI models to be shut down if it is found to pose a “catastrophic risk.”

“States retain the ability to pass and enforce laws regarding every other aspect of AI safety, including child and artist protections, civil rights and other important laws. None of that is stopped by this bill!” said Trahan.

Trahan spokesperson Francis Grubar on Friday pointed to Trahan’s opposition to a Republican proposal to impose a 10-year moratorium on state AI laws.

“The notion that she would turn around and introduce legislation to not only follow through on the same partisan proposal she vocally opposed, but also go even further by making the ban on AI laws permanent, is just ridiculous. The language simply does not exist in the FRONTIER Act, and claims that it does are flat out wrong. Unfortunately, the organization (QuitGPT) that spearheaded this effort has not responded to offers to meet with our office regarding their claims,” said Grubar.

“Despite this, Lori and our team always appreciate folks here in the Third District making their voices heard on vitally important issues like AI safety, and we’re glad that we share their perspective opposing a ban on state AI laws,” Grubar continued. “Lori will continue to work on federal legislation to mitigate the greatest risks posed by frontier AI models, including catastrophic risks to our financial system, energy grid, infrastructure, and more, and she will continue to work across the aisle to advance those policies when she finds a Republican counterpart willing to work in good faith on this issue.”

Caleb Knapp, the senior policy manager at the Alliance for Secure AI, a nonprofit aimed at educating people about the risks posed by the rise of AI, worked with Trahan and the other congressional supporters of the bill to refine it after much of its framework was originally introduced through the Great American AI Act.

“That first stage we thought was a much weaker bill, overly vague, a lot worse,” Knapp said in a phone call Friday.

Knapp said it is likely this Frontier Act sees some tweaks still as the bill’s sponsors gain wider support for it, but echoed Trahan in saying the bill only preempts states from narrow categories of new regulations while also imposing regulations at the federal level.

“We are still working to make the language as strong as possible to ensure it doesn’t accidentally cover other areas,” said Knapp.

After delivering the petition, the group of protesters walked back to Kerouac Park where they had gathered earlier. Among them at the park was Lowell City Councilor Sean McDonough, who said giving large AI firms “even more leeway at the state and local level” is not the direction to go in.

“We need access to more tools and more support in dealing with companies like this, both in terms of the infrastructure buildout and the products that the clients of that data center are pushing,” said McDonough. “Having federal legislation that could open the door to tying our hands behind our back and taking away the toolbox is the exact opposite of what we need to be doing.”

One of the issues McDonough sees with the preemption section of the bill is that it relies on “broad definitions,” which he said usually “benefits whoever has the most expensive lawyer.”

“One issue is how you define what it means to develop, and the bill does say that development could include things like training and fine tuning, which is done at a much smaller scale. The model is built in a particularly large facility,” said McDonough. “On a state level, opening the door to preemption for these multibillion-dollar tech titans, I hate to do the slippery slope, but this is not an industry that needs less regulation.”

Also with the protesters was Jake Fortes, who has been a prominent critic toward data centers, which are needed to power AI models, in particular that of the Markley Group near his home in Lowell’s Sacred Heart neighborhood. Fortes said his concern with the legislation is that it “takes the right away from state and local authorities to safely site [data centers] so that something like Markley could happen all over again.

“I am not down for that. That needs to come out of the legislation,” said Fortes.