Every AI Policy Ever · open in the interactive register
Partially in forceBinding · Law
Record US-CT-01 · Public Act No. 26-15 (Substitute Senate Bill No. 5, 2026)
Omnibus AI and online-safety law. Frontier developers (over 10^26 operations) may not adopt rules or contracts allowing retaliation against employees reporting catastrophic-risk dangers (over 50 deaths or USD 1 billion in damage from expert CBRN weapons help or unsupervised cyberattacks), and large ones need anonymous internal reporting. AI companion operators must detect suicide and self-harm risk, refer users to 988, disclose that the companion is not human and shield minors from harmful or manipulative conduct. It also requires notices on AI in employment decisions, provenance data in AI-generated media, state agencies' AI impact assessments and limits on minors' algorithmic feeds.
Approved by the Governor on 27 May 2026; the Governor's press release announcing the signing is dated 2 June 2026, and a sponsor calls the law the C.A.R.T. Act. Effective dates are staggered: sections 16, 28, 29, 32 and 35 on passage; the AI Academy, the legislative AI working group and several education and workforce sections on 1 July 2026; the subscription, frontier-developer, employment-decision, anti-discrimination, provenance and state-agency sections on 1 October 2026, with large frontier developers' internal reporting due by 1 January 2027 and the duties on automated employment-related decision technology applying to technology deployed from 1 October 2027; the AI companion sections 4 to 6 on 1 January 2027; the regulatory sandbox plan (section 3), the Department of Consumer Protection's pilot of independent verification organisations for AI models (section 33, until 30 June 2030) and the Chief Data Officer changes on 1 July 2027; and the minors' social media section 39 on 1 January 2028. Violations of the subscription, AI companion, employment-decision and provenance sections are unfair or deceptive trade practices enforced solely by the Attorney General, and sections 1, 5, 12 and 15 rule out a private right of action; a frontier developer is liable for civil penalties of up to USD 1,000 per violation of section 2. Under sections 13 and 14 the use of automated employment-related decision technology is no defence to a discrimination complaint.
Open the official source https://www.cga.ct.gov/2026/ACT/PA/PDF/2026PA-00015-R00SB-00005-PA.PDF
Secondary source: https://portal.ct.gov/governor/news/press-releases/2026/06-2026/governor-lamont-signs-legislation-establishing-youth-online-safety-protections
Archived. Last attempt 2026-10-17 23:06:10 UTC
California SB 53 (TFAIA) California (US) · In force
New York RAISE Act New York (US) · Adopted, not yet in force
Washington AI companion chatbot law (HB 2225) Washington (US) · Adopted, not yet in force
California SB 243 (companion chatbots) California (US) · In force
Oregon AI chatbot safety law (SB 1546) Oregon (US) · Adopted, not yet in force
Illinois AI Safety Measures Act (SB 315) Illinois (US) · Adopted, not yet in force
Georgia AI companion chatbot law (SB 540) Georgia (US) · Adopted, not yet in force
Rhode Island AI Companion Models Act (S 2195) Rhode Island (US) · Adopted, not yet in force
Connecticut General Assembly (2026 Session); signed by Governor Ned Lamont. An Act Concerning Online Safety (Public Act 26-15; Substitute Senate Bill 5) [Public Act No. 26-15 (Substitute Senate Bill No. 5, 2026)]. 2026-05-27. https://www.cga.ct.gov/2026/ACT/PA/PDF/2026PA-00015-R00SB-00005-PA.PDF. Status: Partially in force; verified 2026-10-18 (level A). Local copy archive/US-CT-01/2026-10-17/official.pdf retrieved 2026-10-17 22:16:08 UTC, SHA-256 98052c119778bff6d594a94c4a9f927f75212a2d814cf27afcf30d9274c11fbf. Every AI Policy Ever, version 1.18 (snapshot 2026-10-18), record US-CT-01.
Know of an amendment, a status change or an error? Report news or an error on GitHub; the form opens with this record filled in.