Connecticut Senate Approves Major AI Legislation

(TNS) — Amid opposition from a national trade organization and skepticism from Gov. Ned Lamont, the Democratic majority in the state Senate on Wednesday night approved legislation that would regulate artificial intelligence and criminalize deepfake porn and false political messaging. It also aims to protect consumers against racial, gender and economic bias.

The bill was approved after nearly four-and-a-half hours of debate in a party line, 24-12 vote.

The debate culminated a two-year effort by state Sen. James Maroney, D-Milford, co-chairman of the General Law Committee, who helped lead a statewide working group on the issue and is participating in national efforts as states grapple with the emerging technology. In recent weeks, against opposition from the Virginia-based Consumer Technology Association and Lamont’s worries that tight regulations could stifle the nascent industry, Maroney continued to rewrite the legislation up until a final draft Wednesday morning.

“A lot of us don’t have faith in the federal government to act,” Maroney said shortly before the final vote. He stressed that federal legislature on internet-related issues haven’t been adopted by Congress since 1998. “We know we have to approach AI differently. There is so much more change to come. If nothing changes, nothing changes.”

Several Republicans praised Maroney for embracing the issue, but said they wanted to protect businesses and think it is too soon to act. Part of the bill would create an online academy through Charter Oak State College for the public to learn about the fast-moving AI technology.

This week, Maroney deleted a section of the legislation that would have required AI developers to provide detailed technical information to the attorney general or the state Department of Consumer Protection. His interest in AI, he said, was a natural extension of a previous three-year effort to create a data-privacy bill in 2022.

Sen. Eric Berthel of Watertown, a ranking Republican on the budget-writing Appropriations Committee, first demanded that the bill be referred to the committee because of financial implications, suspending action for several minutes, as Democratic Senate leaders huddled nearby in the historic third-floor chamber. Berthel soon withdrew the motion, but warned that costs of the legislation might remain and may tie future legislatures to the program.

On the fiscal implications, Maroney said that the Connecticut Academy of Science and Engineering is offering free assistance and associated grant programs, while earlier versions of the bill were deleted or the funding will include federal pandemic relief money.

State Sen. Paul Cicarella of North Haven, a ranking Republican on the General Law Committee, reviewed the latest draft with Maroney, asking how the bill would regulate AI developers working out of state. “It would be enforced on any company that would want to do business in the state of Connecticut,” Maroney said.

“This is going to make it more challenging and cumbersome for businesses to do business in the state of Connecticut, and possibly more expensive,” Cicarella said, stressing that many small businesses might not even know the AI they use might make them liable for violating state law. “I want to make sure that no one gets in trouble for people who don’t know what they are doing,” he said.

Maroney responded that transparency, in which developers would be required to label applications that use AI, is an important part of the consumer protections included in the bill. He stressed that healthcare is mostly exempted from the legislation.

The legislation targets potential AI discrimination based on age, religion, gender identification, religion and other protected classes impacted in housing, education, finance, employment, criminal justice, education, government jobs and insurance coverage. “A lot of this is going to come down to contract law,” Maroney said. Part of the bill would take effect as soon as July 1, with the rest effective in July, 2025 and as late as February 2026.

“AI is all around us,” said Sen. Tony Hwang, R-Fairfield, who has served on study groups with Maroney. “I’m trying to understand this better.” Hwang said that health care professionals have expressed concerns to him about liability issues over the collection and retention of data. Maroney responded that there are provisions to protect well-intentioned people, companies and government.

“I don’t think anyone has a first grasp on this,” said Senate Minority Leader Stephen Harding who complained that lawmakers had only minutes to read and understand “the most-complex bill” facing the General Assembly this year. “I feel this legislation is filled with potential unintended consequences.”

State Sen. Saud Anwar, D-South Windsor, a physician, warned that the AI industry is changing so fast that lawmakers will have to revisit the issue every year going forward. “I would even say that not doing a bill on this topic would be malpractice on the part of the legislative body,” said Awar. “In 2024, if the state of Connecticut does not create processes to protect the people of the state of Connecticut from artificial intelligence and how it can be used in a negative way, we are not being responsive.”

During the debate, Julia Bergman, spokesperson for Lamont, was noncommittal on what is a top priority for Senate Democrats with the title of S.B. 2. “The governor appreciates all the hard work Sen. Maroney has put into Senate Bill 2,” Bergman said in an afternoon statement. “We are supportive of efforts to increase disclosure of the use of AI in elections, and to close loopholes around intimate images. However, the governor remains concerned that this is a fast moving space, and that we need to make sure we do this right and don’t stymie innovation.”

“He has been extraordinarily transparent about the entire enterprise, despite its complexity,” said Senate President Pro Tempore Martin Looney of Maroney’s work. “It’s a critically important bill because we believe that in an area like this, where the influence of AI is going to be so pervasive in ways that may not even be known yet, that the danger of failing to adopt adequate and reasonable regulation is greater than any potential danger of over-regulation.”

Senate Majority Leader Bob Duff, whom Maroney earlier credited with getting him first involved in data privacy, then AI, summed up the debate at about 7:35, said that AI is evolving “by the minute” and is complex. “When he speaks on AI, people want to listen,” Duff said, noting that Maroney has been a national voice for lawmakers looking at AI at a time when social media is unregulated and data privacy is left to the states. He warned that AI companies are currently scraping the entire internet, collecting content, despite copyrights.

“We should make sure that it is good for everybody, that we take out discrimination, that we take out biases,” Duff said, warning that at this moment, Connecticut can lead the nation in the growth of the AI industry and still protect residents.

© 2024 The News-Times (Danbury, Conn.). Distributed by Tribune Content Agency, LLC.