Connecticut Bans AI-Only Health Insurance Claim Denials, Requires Human Review

Connecticut Comptroller Sean Scanlon announced new regulations on September 16, 2026, governing how health insurance providers can use artificial intelligence when administering state employee health plans. The rules take effect January 1, 2027.

The regulations apply to approximately 270,000 enrollees, including state employees and participants in the Connecticut Partnership Plan for municipal and public-sector workers. They were developed through negotiations with the insurance carriers managing the state's health benefits, including Anthem, Cigna, Aetna, and Caremark.

The policy establishes five requirements. First, AI systems cannot act as the sole decision-maker for adverse determinations. All such decisions must be reviewed by a human. Second, carriers may not use AI or predictive models as the exclusive basis for down-coding claims, reducing provider payments, or altering billing codes without human intervention.

Third, member health data is protected from being used to train, develop, or support other AI models. Fourth, carriers and providers must disclose when AI is materially assisting with or recommending a health service or benefit. Fifth, AI systems must be validated for accuracy, consistency, and fairness against historical data, and carriers must disclose their governance and audit procedures to the Comptroller's office.

Because state authority is limited to "fully insured" plans, the rules do not currently reach the majority of residents covered by "self-funded" plans, which are regulated at the federal level.

Comptroller Scanlon said he intends to seek legislation in 2027 to extend similar protections to all state-regulated health plans. Thirteen other states have already enacted various forms of restrictions on the use of AI in health insurance decision-making.