Connecticut Passes Comprehensive AI Regulation Bill SB5
On May 1, Connecticut SB5 received approval from both chambers of the legislature and is now pending the governor’s signature, positioning it as one of the most comprehensive state-level AI regulations in the United States.
Summary
- Connecticut SB5 passed in the House with a vote of 131-17 and in the Senate with a 32-4 vote on May 1; Governor Lamont has indicated he will sign the bill.
- This legislation addresses issues related to AI companions, the necessity for transparency in synthetic media, automated employment decision tools, and regulations for frontier model developers, featuring staggered implementation dates starting in October 2026.
- Despite an executive order from the Trump administration opposing stringent state-level AI regulations, this law will take effect, establishing Connecticut as a state that resists federal overreach.
Connecticut SB5 was approved on May 1, emerging as one of the broadest state AI regulations nationwide. The House demonstrated bipartisan support with a vote of 131-17, followed by the Senate with a 32-4 vote. Governor Ned Lamont affirmed his intention to sign the bill, officially named the Connecticut Artificial Intelligence Responsibility and Transparency Act.
The legislation encompasses areas such as AI companions, automated employment decision-making tools, monitoring of synthetic media origins, and regulations affecting frontier model developers crossing defined thresholds.
The initial implementation date is set for October 1, 2026, with most provisions enforceable only by the Attorney General as indicators of unfair or deceptive practices—therefore allowing no private right to file a lawsuit.
Requirements of the Law
Under SB5, employers must disclose when automated tools are utilized in hiring practices, and companies cannot rely on these tools as a defense against discrimination claims. These employment-related stipulations will be enforced starting October 1, 2026.
Rules concerning AI companions, including chatbots designed to foster emotional connections, will come into effect beginning January 2027. Generative AI systems with over one million users are mandated to adhere to C2PA-aligned standards for provenance data.
Developers of frontier models should establish internal safety measures for AI and protect whistleblowers reporting safety concerns. As noted by crypto.news, the regulation of AI companions is gaining traction across multiple U.S. states in 2026, particularly following lawsuits in Pennsylvania and Kentucky regarding harms caused by chatbots.
Collision with Federal Regulations
Connecticut joins California, Colorado, and other states in enacting AI-specific legislation, pushing back against Trump’s executive order aimed at curtailing what are seen as excessive state regulations.
SB5 introduces a regulatory sandbox and a working group, with an initial meeting slated for August 31, 2026, to aid in implementation. As reported by crypto.news, federal agencies are simultaneously employing AI tools to navigate regulatory gaps, resulting in a complicated environment for companies operating across state lines.
Attorney General William Tong mentioned that his advisory to businesses in February 2026 indicates his office considers AI regulation within its scope. SB5 equips his office with advanced, specialized tools to support that stance.