Connecticut's sweeping AI regulation clears the legislature with bipartisan supermajorities, setting the state apart as federal AI governance stalls in Washington.
Connecticut's House of Representatives voted 131-17 on Friday to approve Senate Bill 5, a comprehensive artificial intelligence regulation package, sending it to Gov. Ned Lamont for signature. The Senate had cleared it 32-4 earlier. The lopsided margins are notable for a bill that spent years stalled in the legislature and arrives as the Trump administration presses states to defer to federal authority on artificial intelligence.
The legislation covers AI systems' effects on businesses, workers, and minors - a scope that drove three hours of floor debate. Both parties closed ranks behind it. Rep. David Rutigliano, a Republican from Trumbull, framed the bill as guardrails rather than a brake: "We're not stifling innovation, we're not doing anything to stifle economic development."
Rep. Roland Lemar, a Democrat from New Haven who co-chairs the General Law Committee, opened debate with a broader warning. "There is no longer doubt that the nature of work, the nature of life, is going to change rapidly with the continued evolution of AI," he said. "This is about protecting people without stopping that innovation."
The regulatory push
Connecticut has tried this before. Last year's bill cleared the Senate after a last-minute deal, then collapsed when Lamont threatened a veto, citing concerns that aggressive rules would push businesses to neighboring states. This year's measure appears to have threaded that needle.
What Senate Bill 5 specifically requires - thresholds for covered systems, disclosure obligations, enforcement mechanisms - was debated extensively but not fully documented in records available before publication. Legislators repeatedly raised the challenge of regulating a technology that keeps changing beneath any regulatory framework, according to CT Mirror. The bill's protections for children and teens drew particular attention during Friday's session.
The federal backdrop sharpens the stakes. The Trump White House has intervened directly in decisions about which AI systems can reach the market. GovInfoSecurity reported this week that the administration blocked an expansion of access to Anthropic's Mythos model, restricted over national security concerns, fitting a clear pattern of centralizing AI oversight in Washington. State-level regulation cuts against that approach.
Connecticut's vote positions it alongside California and Colorado as states willing to act without waiting for Congress, which has yet to pass any comprehensive federal artificial intelligence law.
Context and implications
The timing is not coincidental. AI deployment is accelerating faster than any prior technology platform. OpenAI now generates $2 billion in monthly revenue and expects to reach one billion weekly active users. The EU's artificial intelligence act is moving into enforcement. The absence of a U.S. federal equivalent has left states to fill the vacuum on their own.
For companies doing business in Connecticut, the immediate question is compliance cost. A patchwork of state rules, each with distinct thresholds and enforcement regimes, is precisely the scenario the tech industry has lobbied hardest to prevent. The EU's experience shows that even carefully drafted frameworks generate years of interpretive uncertainty before firms know what the rules actually demand of them.
Investment continues regardless of the regulatory climate. Analytics Insight reported this week that Ineffable Intelligence, a new lab founded by former Google DeepMind researcher David Silver, raised a $1.1 billion seed round at a $5.1 billion valuation, the largest seed round on record in Europe. Regulatory friction has not slowed the capital.
The decisive variable is now Lamont's signature. A governor who threatened a veto last year is about to tell Connecticut businesses and the rest of the country watching state AI policy exactly where he stands.
---
FAQ
What does Connecticut's AI bill regulate?
Senate Bill 5 targets AI systems deployed in Connecticut, with protections focused on consumers, businesses, and minors. Full disclosure requirements and enforcement details were still being clarified publicly as of the vote.
Why did Connecticut's previous AI bill fail?
Gov. Ned Lamont threatened a veto over business competitiveness concerns. This year's legislation appears to have addressed enough of those objections to advance with strong bipartisan margins in both chambers.
What is the Trump administration's position on state AI regulation?
The Trump White House has pushed to centralize AI oversight federally and has directly intervened in decisions about which AI models reach the market. It has not publicly supported state-level frameworks.
How does Connecticut's law compare to the EU's artificial intelligence act?
The EU's AI Act is the most comprehensive framework globally and is now entering enforcement. Connecticut's bill covers similar ground at the state level, but in a legal system without federal preemption, companies face the prospect of conflicting rules across state lines.