Several parts of Connecticut's new AI law take effect October 1

Connecticut’s Public Act 26-15, passed as Senate Bill 5 and approved by the governor on May 27, 2026, takes effect in stages. Several sections began on October 1, 2026.

From October 1:

• AI subscriptions. A company selling an AI tool by subscription to Connecticut consumers must give written notice of the key terms, including usage limits and whether it can cut features, and get the customer’s written acceptance before charging or renewing.
• Staff at frontier developers. Companies training very large models, using more than 10^26 computing operations, may not punish covered employees who report a substantial danger to public health or safety from a “catastrophic risk.” The law defines this as a risk that a model materially contributes to the death or serious injury of more than 50 people, or over $1 billion in property damage or loss, in a single incident, for example by helping to make a weapon or by carrying out a cyberattack. Developers with over $500 million in annual revenue must set up anonymous internal reporting by January 1, 2027.
• Layoffs. Employers filing a federal WARN Act layoff notice must tell the state Labor Department whether the layoffs relate to their use of AI or another technological change.
• State agencies. According to the state’s Office of Legislative Research, agencies may not use AI for public assistance benefits, or for functions with a material impact on people’s rights or safety, unless they follow state standards.

Other parts come later. Rules for AI companions start January 1, 2027, and notice duties for AI hiring tools apply to tools deployed from October 1, 2027.

Why it matters

AI plans can cap usage or let the provider reduce features. Connecticut now requires those terms to be spelled out in writing before a customer pays or renews. The What AI Costs course explains how AI plans and prices are set and what you are paying for.