Connecticut Artificial Intelligence Responsibility and Transparency Act (CART Act, Public Act 26-15)

Properties

jurisdiction: Connecticut
level: state
domain: general
status: in-effect
scope: Omnibus act (Substitute Senate Bill No. 5) covering: employer use of automated employment decision tools (disclosure duties, mass-layoff AI disclosure), AI companion chatbots serving minors, frontier foundation-model developers (training compute > 10^26 operations) with whistleblower protections, and generative-AI content provenance requirements. Also contains youth online-safety and workforce-development provisions that are not AI-specific.
obligations: Employers must disclose AI involvement in personnel/layoff decisions; frontier developers must implement whistleblower protections (heightened obligations for 'large frontier developers'); AI companion operators serving known minors must meet protective standards; generative-AI provenance requirements for covered providers.
effectiveDate: Phased: subscription transparency, whistleblower protections, provenance and AEDT framework from 2026-10-01; AI companion requirements from 2027-01-01; AEDT disclosure duties for deployments from 2027-10-01; minor social-media protections from 2028-01-01
enactedDate: 2026-05-27
penalties: Most operative provisions (AI subscription terms, AI companion chatbots, employment automated decision tools, generative-AI provenance) are enforced solely by the Connecticut Attorney General as unfair or deceptive trade practices under CUTPA, with no private right of action; CUTPA civil penalties under general law are up to $5,000 per willful violation (Conn. Gen. Stat. 42-110o). For employment automated decision tool violations occurring on or before December 31, 2027, the Attorney General may issue a notice of violation and the person then has 60 days to cure before an action. Frontier developers that violate the whistleblower provisions are liable to the state for a civil penalty not exceeding $1,000 per violation, collected by the Attorney General in Hartford superior court. The minor social-media provisions (section 39, effective January 1, 2028) are deemed unfair or deceptive trade practices without the Attorney-General-only language.
appliesTo: Employers using AI in employment decisions; frontier AI model developers/trainers; AI companion chatbot operators; generative-AI content providers operating in or reaching Connecticut
officialUrl: https://www.cga.ct.gov/2026/act/pa/pdf/2026PA-00015-R00SB-00005-PA.pdf
notes: Official CGA text confirms Public Act No. 26-15 (Substitute Senate Bill No. 5), 'An Act Concerning Online Safety', with 'Governor's Action: Approved May 27, 2026' (CORRECTION 2026-10-25: enactedDate was recorded as 2026-06-02, the date of the Governor's press release). Section effective dates in the act text: some sections from passage and July 1, 2026; most sections (including AEDT framework sections 8 to 11, provenance and whistleblower provisions) October 1, 2026; AI companion sections 4 to 6 January 1, 2027; minor social-media protections (section 39) January 1, 2028; AEDT deployer disclosure duties apply to deployments on or after October 1, 2027. Because the October 1, 2026 tranche has now arrived, status moves from enacted-pending to in-effect (phased). Governor's office press release: https://portal.ct.gov/governor/news/press-releases/2026/06-2026/governor-lamont-signs-legislation-establishing-youth-online-safety-protections Domain detail: cross-sector (employment AEDT, AI companions/chatbots, frontier models, generative-AI provenance).

Properties

jurisdiction

Connecticut

level

state

domain

general

status

in-effect

scope

Omnibus act (Substitute Senate Bill No. 5) covering: employer use of automated employment decision tools (disclosure duties, mass-layoff AI disclosure), AI companion chatbots serving minors, frontier foundation-model developers (training compute > 10^26 operations) with whistleblower protections, and generative-AI content provenance requirements. Also contains youth online-safety and workforce-development provisions that are not AI-specific.

obligations

Employers must disclose AI involvement in personnel/layoff decisions; frontier developers must implement whistleblower protections (heightened obligations for 'large frontier developers'); AI companion operators serving known minors must meet protective standards; generative-AI provenance requirements for covered providers.

effectiveDate

Phased: subscription transparency, whistleblower protections, provenance and AEDT framework from 2026-10-01; AI companion requirements from 2027-01-01; AEDT disclosure duties for deployments from 2027-10-01; minor social-media protections from 2028-01-01

enactedDate

2026-05-27

penalties

Most operative provisions (AI subscription terms, AI companion chatbots, employment automated decision tools, generative-AI provenance) are enforced solely by the Connecticut Attorney General as unfair or deceptive trade practices under CUTPA, with no private right of action; CUTPA civil penalties under general law are up to $5,000 per willful violation (Conn. Gen. Stat. 42-110o). For employment automated decision tool violations occurring on or before December 31, 2027, the Attorney General may issue a notice of violation and the person then has 60 days to cure before an action. Frontier developers that violate the whistleblower provisions are liable to the state for a civil penalty not exceeding $1,000 per violation, collected by the Attorney General in Hartford superior court. The minor social-media provisions (section 39, effective January 1, 2028) are deemed unfair or deceptive trade practices without the Attorney-General-only language.

appliesTo

Employers using AI in employment decisions; frontier AI model developers/trainers; AI companion chatbot operators; generative-AI content providers operating in or reaching Connecticut

officialUrl

https://www.cga.ct.gov/2026/act/pa/pdf/2026PA-00015-R00SB-00005-PA.pdf

notes

Official CGA text confirms Public Act No. 26-15 (Substitute Senate Bill No. 5), 'An Act Concerning Online Safety', with 'Governor's Action: Approved May 27, 2026' (CORRECTION 2026-10-25: enactedDate was recorded as 2026-06-02, the date of the Governor's press release). Section effective dates in the act text: some sections from passage and July 1, 2026; most sections (including AEDT framework sections 8 to 11, provenance and whistleblower provisions) October 1, 2026; AI companion sections 4 to 6 January 1, 2027; minor social-media protections (section 39) January 1, 2028; AEDT deployer disclosure duties apply to deployments on or after October 1, 2027. Because the October 1, 2026 tranche has now arrived, status moves from enacted-pending to in-effect (phased). Governor's office press release: https://portal.ct.gov/governor/news/press-releases/2026/06-2026/governor-lamont-signs-legislation-establishing-youth-online-safety-protections Domain detail: cross-sector (employment AEDT, AI companions/chatbots, frontier models, generative-AI provenance).