The bill, lawmakers said, will establish Connecticut as a national leader in oversight of a major industry that involves billions of dollars and is a large driver on the stock market. The legislation creates a regulatory framework at the same time that lawmakers say they want businesses to capitalize on the technological revolution.
After more than 2 1/2 hours of debate, the state House of Representatives voted 131-17 last week with all negative votes cast by Republicans. The Senate had previously approved the bill by 32-4 with Republican Sens. Rob Sampson of Wolcott, Tony Hwang of Fairfield, Eric Berthel of Watertown, and Ryan Fazio of Greenwich against the measure.
The legislative compromise marks a sharp change from recent years when Gov. Ned Lamont threatened to veto previous bills because he feared that too much regulation would hinder growth of the industry and prevent high-technology companies from growing in Connecticut or moving to the state.
Lamont’s new spokeswoman, Cathryn Vaulman, said that Lamont favors the bill with “commonsense protections” and will sign it into law.
“Governor Lamont made it a priority this session to fight for protections for Connecticut residents — especially children — from serious threats posed by emerging technology,” she said. “Parents should be in control of aspects of social media and AI that carry real risks for children’s mental health. Workers should be able to benefit from greater efficiency on the job without fearing discrimination or displacement by AI.”
Known as Senate Bill 5, the wide-ranging, 71-page bill would create oversight committees, adopt workforce development programs and attempt to prevent AI from discriminating in hiring and sorting through resumes in the employment process. Connecticut small businesses would not be held liable if a third-party vendor in the hiring process uses AI, lawmakers said.
Rep. Hubert Delany, a Stamford Democrat who co-chairs the legislature’s AI caucus with Sen. James Maroney of Milford, said the final bill was written after “years of hard work” on technology that is “perhaps the most powerful and transformative of our generation” and will continue changing.
“Right now, without standards and guidelines … consumers have little guidance or trust with this technology,” Delany said. “Could you imagine flight without traffic control? … This legislation builds highways to progress. … The horse is artificial intelligence, and it is high time that we hitched the cart to it.”
Rep. Roland LeMar, a New Haven Democrat who co-chairs the general law committee that oversees AI, said that the legislature will continue looking at AI as the technology evolves.
“We will support innovation and business growth,” LeMar said. “We have the opportunity to lead — not follow — on the defining technology of our time. We need to act now. … It ensures that AI works for people, not against them. … We’ll be talking about AI and regulatory structures for the rest of our lives.”
Rep. Dave Rutigliano, a Trumbull Republican who serves as the ranking member on the general law committee, said lawmakers took pains to strike a delicate balance.
“We’re trying to walk a fine line between consumer protection, protecting our young people and not stifling innovation and industry,” Rutigliano said on the House floor. “It’s illegal for AI to discriminate” in hiring and employment practices.
Overall, Rutigliano said the legislation is “a good first step” that will help consumers.
“The time has come,” Rutigliano said. “We’re not stifling innovation. … Let’s be clear. If you’re using chatbot and it’s telling you to do things that aren’t right, it shouldn’t be doing that. … I think it’s OK for us to start a regulatory environment. … It’s certainly not perfect, but it is a start.”
AI REFLECTING HUMAN ACTIVITY
Artificial intelligence has improved to the point where consumers sometimes do not know whether they are speaking to a real human being when they call an 800 number for customer service.
As a result, a nonpartisan bill summary states that companies must “prevent the AI companion from claiming to be a human being, including when a user asks the companion if it is a human being, or generating outputs that refute or conflict with any disclosure that the companion is not a human being.”
In the same way, minors and others sometimes develop interaction with an “AI companion,” that often responds in the same way as a human. The companion not only provides human-like responses but “can sustain a relationship across multiple interactions,” a bill summary states.
The companion must use “evidence-based methods to detect any user expression clearly showing a risk of suicide, self-harm, or imminent physical violence, and implement prevention measures to prevent the AI companion from generating output that encourages this behavior,” the summary states. “If the companion detects any of these expressions, it must refer the user to appropriate mental health evaluation and treatment resources, including the 9-8-8 National Suicide Prevention Lifeline.”
Rep. Bill Buckbee, a New Milford Republican, said AI companions can be helpful for senior citizens “who are alone and need somebody to talk to.”
The bill also allows state Attorney General William Tong to provide more oversight regarding AI since his office primarily enforces the law.
Tong’s office said that the bill “requires chat bot operators to adopt protocols to detect self-harm, to provide notices that users are communicating with an AI companion, and to require chat bot operators to prohibit access to minors if the program is capable of encouraging self-harm and offering mental health services, among other provisions.”
The bill also requires “disclosure requirements related to use of automated decision technology in employment decisions, requires that state notifications given prior to mass layoffs and plant closings note when such actions are related to use of AI or other technology, and requires the state to study and develop a plan to address AI-related employment impacts, among other provisions.”
CONCERNS ABOUT JOBS
A major concern about AI is how it will impact the workforce.
Many predictions have been made, for example, that there will be a reduced need for paralegals to conduct legal research and fewer radiologists because AI can read X-rays and MRIs. But radiologists are still employed because some insurance companies will refuse to pay claims unless they are overseen by radiologists and cannot be done exclusively by AI, officials said.
In addition, AI has generated legal briefs that cited incorrect legal cases that ended up in court.
Lamont’s office says that AI is “transforming workplaces across Connecticut and the nation” as the demand for skills in the field is increasing.
“Since August 2024, nearly 11,000 job postings have required AI skills — a 40% increase from the year prior,” Lamont’s office said earlier this year. “Today, one in 52 jobs in the state list AI skill requirements, up from one in 70 a year ago. The demand is even stronger in roles requiring an associate or bachelor’s degree, where one in 23 jobs now call for AI expertise, spanning nearly every industry and occupation.”
FLOOR DEBATE
Rep. Gary Turco, a Newington Democrat who is among the legislature’s leading authorities on artificial intelligence, handled much of the questions regarding the bill on the House floor.
“AI is rapidly growing,” Turco said. “This is the wave of the future.”
With some consumers increasingly trying to get their medical information on the Internet, Turco said the bill provides safeguards for physical and mental health.
“The platform has to tell them that they are not a licensed medical professional,” Turco said. “There are very strict restrictions in this bill.”
On a highly complicated issue with detailed explanations, lawmakers said they sometimes needed to explain the bill at “the fourth grade level” so that newcomers to the issue could understand it.
Lawmakers said that AI will be incorporated into the curriculum in computer science classes.
Seeking to protect children, the bill says that chatbots “cannot encourage any romantic” interactions in a measure that can be enforced by the state attorney general, lawmakers said.
Rep. Geraldo Reyes, a Waterbury Democrat who supported the bill, said he wants to make sure that young people are protected.
“I consider myself a complete dinosaur when it comes to AI,” Reyes said. “I can tell by the way my grandchildren interact with me with AI.”
But Rep. Thomas O’Dea, a New Canaan Republican, said he has concerns about the impact on the workforce.
“Look, AI frankly petrifies me,” O’Dea said.
Rep. Joe Hoxha, a Bristol Republican, said, “I’m not sure that those who created this modern-day Frankenstein know how to control it. … This is really a new frontier. … Sometimes fiction is a lot closer to reality than we might think. … This monster that we humans have created will take us over one day if we do not get it under control.”
Rep. Tina Courpas, a Greenwich Republican, said she was concerned that the bill “restrains innovation,” but she was open-minded to learn more about AI.
Rep. Christie Carpino, a Cromwell Republican, said, “AI is already here. It’s not coming. … It’s not going to protect our residents if I vote no. … Chatbots scare me when they interact with unsophisticated residents. … Do I think the federal government — I can’t believe I’m saying this — should take some action? I do. … If we do nothing, it is a disservice to the people we represent.”
DEEPFAKES
For years, the artificial intelligence bill had a key provision regarding “deepfake porn” that involves creating videos or photographs — often of famous people — that would show their recognizable head but attached to another’s body in compromising positions. The deepfake porn provisions were not passed into law when the overall AI bill was shelved for the year.
In this session, deepfakes were placed separately in House Bill 5312, which was known as the “Take It Down Act.” The measure allows civil lawsuits by the state attorney general, as well a private right of action that allows Connecticut residents to sue if they are “victims of unlawful dissemination of a synthetically created intimate image.”
The bill passed by 148-0 with three members absent in the House.
LONG HISTORY IN CT
Concerns about artificial intelligence have a long history.
In 2007, famed actor Paul Newman of Westport traveled to the state Capitol complex in Hartford and testified to the judiciary committee for a bill that would have prevented the unauthorized digital distortion of anyone’s image and create “a right of publicity” in the digital age. Newman was way ahead of his time 19 years ago, and he told lawmakers that technology had improved so quickly that he feared that film clips from his performances could be digitally extracted for use in other films or commercials. Newman, a Hollywood giant, died in 2008 at the age of 83.
The unauthorized use of images was not difficult to accomplish and could be done by “the average guy in the basement,” Newman told The Courant in an interview after his testimony.
The 2007 bill, which did not pass, would have prevented any distortion that would “cause the individual to speak or appear to speak words that the person did not speak or place the individual or appear to place the individual in a place or circumstance in which the individual did not agree to be placed.”
The current technology has gone far beyond the 1994 classic film “Forrest Gump,” in which lead actor Tom Hanks’ character appeared in scenes shot decades earlier, meeting with President Lyndon Johnson and appearing with former Beatle John Lennon on The Dick Cavett Show. But both Cavett and Lennon’s widow, Yoko Ono, gave permission for the latter scene, officials said.
©2026 Hartford Courant, Distributed by Tribune Content Agency, LLC.
