Connecticut mandates AI disclosures Oct 1

Connecticut has moved ahead with one of the broader state AI laws enacted this year, with the first major compliance deadline set for October 1, 2026. Governor Ned Lamont signed SB 5 in late May, creating new requirements across consumer-facing AI, synthetic content, employment tools and other uses of artificial intelligence. The law phases in over time, but several of the most immediate obligations start this fall. October 1 matters because the law does not stop at one narrow use case. Connecticut’s bill analysis says that, beginning that day, anyone doing business in the state who deploys an AI system that interacts with consumers must ensure each consumer is told they are interacting with AI. The same date also starts documentation, risk management and disclosure duties for developers, integrators and deployers of certain “high-risk” AI systems. (fpf.org) ### Which Connecticut systems have to disclose AI use first? Section 6 of the bill analysis says consumer-facing AI interactions must be disclosed beginning October 1, 2026. That means the rule is not limited to hiring software; it reaches any business operating in Connecticut that uses AI systems directly with consumers. The law also covers synthetic digital content. (cga.ct.gov) Connecticut’s bill analysis says developers of AI systems capable of generating synthetic digital content must include labels and ensure technical solutions are effective, also beginning October 1, 2026. That creates an early compliance track for provenance and labeling, not just chatbot notices. ### What does the law require for higher-risk AI systems? Connecticut’s Office of Legislative Research analysis says developers, integrators and deployers of high-risk AI systems must use “reasonable care” to protect consumers from known or reasonably foreseeable risks of algorithmic discrimination starting October 1, 2026. Developers must provide deployers with statements describing intended uses and documentation tied to risk mitigation. (cga.ct.gov) Deployers face a longer checklist. The same bill analysis says deployers must implement a risk management policy and program before deployment, complete an impact assessment before deploying or substantially modifying a system, review each deployed system at least annually, and provide policies, assessments and records to the attorney general if relevant to an investigation. (cga.ct.gov) ### How does hiring fit into the Connecticut package? A separate Connecticut bill analysis on employee protections says employers using high-risk AI systems for consequential decisions must have an impact assessment before deployment and give employees information about the system and its use. The analysis says those systems can affect hiring, termination or compensation. (cga.ct.gov) Nixon Peabody and Ogletree summaries of the new law say October 1, 2026 also starts a requirement for employers filing federal WARN notices to tell the Connecticut Department of Labor whether layoffs are related to AI or other technological change. Those summaries describe the law as adding disclosure obligations for employment-related AI use. (cga.ct.gov) ### Why is this landing in a wider AI regulation fight? CNN sued Perplexity in New York federal court on May 28, alleging the company unlawfully copied thousands of CNN stories, videos and images to power its products. Reuters reported that Perplexity responded, “You can’t copyright facts.” (nixonpeabody.com) The White House also released a National Policy Framework for Artificial Intelligence on March 20, 2026, with legislative recommendations for Congress. The White House said it wanted Congress to work toward federal standards while addressing issues including intellectual property, infrastructure and preemption of some state laws. (usnews.com) ### What should companies do before October 1? October 1, 2026 is the first date companies should build toward in Connecticut. Based on the enacted bill analysis, businesses with consumer-facing AI should map where users interact with AI and add disclosures, while companies using higher-risk systems should prepare documentation, impact assessments, risk-management processes and recordkeeping that can be produced in an attorney general investigation. (whitehouse.gov) Connecticut’s own summaries show the law phases in beyond this year, with effective dates extending into January 2028. But the first visible checkpoint is this fall, when disclosure, labeling and oversight duties begin to attach to systems already in use. (fpf.org) (cga.ct.gov)