Additional provisions of Connecticut's artificial intelligence law took effect October 1, requiring employers to disclose whether mass layoffs stem from AI use and mandating that large AI providers label computer-generated content when feasible. The Yankee Institute published the report October 1, noting that state legislative leaders are already drafting a second round of AI regulations before the broadest workplace requirements of the current law begin in 2027. Connecticut could be writing new AI rules before its existing law has fully rolled out.
Starting October 1, employers sending federally required Worker Adjustment and Retraining Notification Act notices to Connecticut's Department of Labor must now reveal whether layoffs connect to AI or other technological shifts. Companies using software to screen job applicants remain liable for discriminatory decisions, unable to blame automated systems. Generative AI systems with over 1 million monthly users that sell subscriptions in Connecticut must outline key terms in writing, including usage limits and access restrictions that customers accept at signup or renewal. State agencies can deploy AI for public assistance or work affecting residents' rights, safety or welfare only after completing impact assessments and following state standards. Additional rules arrive in stages: January 2027 brings safeguards for AI companion chatbots against encouraging self-harm or violence, October 2027 triggers employer notice requirements when AI tools help make hiring or job decisions, and January 2028 limits online platforms from using minors' activity to shape their feeds without parental consent.
Senate President Martin Looney, Senate Majority Leader Bob Duff and Sen. James Maroney praised an Illinois law in July and called for tougher rules on companies building AI systems, aiming for Connecticut action in 2027. CBIA Vice President Christopher Davis backed limiting oversight of powerful AI models to the companies creating them while cautioning against imposing those rules on small businesses using off-the-shelf tools. Labor unions want more than notice: AFSCME Council 4 Deputy Director Zak Leavy urged lawmakers to make AI deployment a mandatory collective bargaining subject, while Connecticut AFL-CIO President Ed Hawthorne proposed requiring bargaining when AI could affect wages, hours, job duties, discipline or other working conditions, including for municipal employees and teachers. Neither proposal became law.
The tension reflects a broader challenge states face as AI adoption accelerates faster than regulation can follow. Connecticut structured its law with staggered deadlines specifically to allow each set of rules time to take effect before the next wave begins, yet legislative leaders are now pushing for additional restrictions targeting AI developers before employers even face the 2027 workplace notice requirements. The report warns that New York's approach offers a cautionary example: that state bars covered public-sector employers from assigning AI work their employees already perform, even without job eliminations, which an analysis from the Manhattan Institute suggests could prevent agencies from using AI to improve public services. The report argues that if software can clear backlogs or deliver faster answers to residents, agencies should be allowed to deploy it, though they must still answer for wrongful denials of benefits rather than blaming the technology.
Lawmakers return in January with a UConn study on AI's effect on Connecticut jobs due just before the session and recommendations from their AI working group due in February. The report concludes that before adding restrictions, lawmakers should be able to name the harm they would prevent and explain what useful work their rules might delay. Connecticut has plenty of experience making people wait on government—it doesn't need an AI law that makes them wait longer.