Connecticut Enacts CART Act to Regulate AI Ecosystem
The new omnibus law establishes a comprehensive regulatory framework for AI developers and applications, moving beyond targeted rules.
Connecticut Governor Ned Lamont signed Senate Bill 5 (SB5) on May 27, 2026, establishing a wide-ranging regulatory framework for artificial intelligence. Now known as the Connecticut Artificial Intelligence Responsibility and Transparency (CART) Act (Public Act 26-15), the law signals a shift toward comprehensive state-level oversight of the AI industry.
Unlike previous legislative trends that focused on single-issue risks, the CART Act is an omnibus law. The statute covers multiple domains simultaneously, including healthcare, employment, and online safety. It imposes distinct obligations across the AI lifecycle, targeting frontier-model developers, generative AI provenance, and the deployment of automated decision tools in employment.
Implementation and Immediate Mandates
The first wave of regulatory duties begins on October 1, 2026. These initial requirements focus on safety protocols for frontier developers and the establishment of whistleblower duties. Additionally, the law mandates generative AI provenance for large-scale systems producing image, audio, and video content to ensure transparency in synthetic media.
Specific protections are also carved out for consumer-facing technology. The Act requires AI companions and chatbots to implement evidence-based measures to detect user expressions regarding self-harm and execute preventative actions to mitigate those risks.
A Shift in Regulatory Strategy
The CART Act arrives as several U.S. states, including Colorado, Utah, and Iowa, tighten their AI restrictions. However, Connecticut's approach differs in scope. While Colorado has focused specifically on automated decision-making, Connecticut has opted for a holistic strategy that addresses the entire ecosystem, from the developers of foundational models to the end-user applications.
Industry Implications
This omnibus approach creates a centralized but complex compliance burden for AI companies operating within the state. By combining sector-specific rules for healthcare and employment with broad safety and transparency mandates, Connecticut is positioning itself as a potential blueprint for other states seeking to move away from fragmented, narrow legislation.
Future Outlook
Industry observers will be watching the October 1 deadline to see how frontier developers adapt their safety and reporting protocols to meet the new state standards. While the broad framework is now law, the specific operational impact on AI companions and the enforcement of provenance requirements for generative media remain the primary points of interest for compliance officers.