Connecticut just put a stake in the ground on two of the thorniest problems of the digital era — kids and social media, and the spread of AI into everyday life — and the throughline in how state officials are talking about it is impatience. Governor Ned Lamont has signed Public Act 26-15 into law, a bipartisan package built jointly by Lamont, Attorney General William Tong, and State Senator James Maroney (D-Milford). Their shared message: with Congress stalled, the states are going to act.
Tong put it about as bluntly as a press release allows, framing the law as a step toward reclaiming parental control from tech platforms he accused of knowing their products are addictive and harmful and not caring as long as the profits keep coming. Lamont struck a more measured note, describing the bill as a balance between protecting children, treating workers fairly, and leaving room for innovation. Maroney, who has become the legislature’s go-to figure on tech policy, cast it as protecting residents while still fostering the industry.
So what does it actually do? The law moves on three fronts.
1. Cracking down on youth social media addiction
This is the part with the most teeth, and the most concrete rules. The law requires social media companies to take a series of steps aimed at anyone under 18.
The headline requirement is age verification with parental consent. Platforms have to verify a user’s age, and if that user turns out to be a minor, the company must get permission from the minor’s parent or legal guardian before the minor can access an addictive algorithmic feed. In other words, the personalized, engagement-optimized feed — the thing the whole business model runs on — is gated behind a parent’s sign-off for kids.
On top of that, the law sets several default protections for minors:
- No late-night notifications. Apps can’t push notifications to minors between 9:00 p.m. and 8:00 a.m.
- A one-hour algorithmic cap. By default, a minor’s access to any part of the platform driven by a personalized algorithm is limited to one hour per day.
- Parental override. Each of these defaults can be loosened or removed — but only with permission from the minor’s parent or guardian.
Finally, the law adds a transparency requirement. Companies must file annual disclosures covering the prior calendar year, reporting figures like how many users obtained parental consent, how many ran on the default settings, and the average time users spent on the platform. That reporting matters because it gives regulators and the public a way to see whether the protections are actually being used or quietly bypassed.
2. Putting guardrails on AI
The second front is artificial intelligence, where the law stakes out two protections.
The first is aimed squarely at a problem that’s been in the news with growing frequency: AI chatbots and self-harm. Under the law, chatbot operators must make reasonable efforts to detect when a user is expressing suicidal ideation or signs of self-harm, and must have a protocol in place to respond with appropriate resources. State officials framed this as especially urgent given how fast chatbots have been deployed to the public without safeguards built in.
The second is about AI in the workplace. Employees have to be notified when AI is being used to make hiring and employment decisions — the kind of automated screening and evaluation that can quietly shape a career without the person ever knowing it happened. The goal is to combat unintended discrimination by at least making the use of these systems visible.
3. Building an AI-ready workforce
The third piece is less about restriction and more about investment — and it’s where the “balance” language from Lamont and Maroney comes through. The law sets up a cluster of programs and partnerships meant to keep Connecticut’s workforce from falling behind:
- AI literacy training for nonprofits and small businesses, so they can adopt the tools without being steamrolled by them.
- An AI Academy, connected to residents through partnerships with the state Departments of Housing and Labor and the Secretary of the State, to teach practical AI skills.
- An AI regulatory sandbox, giving developers a supervised environment to experiment and collaborate while keeping regulators in the loop.
- Government modernization efforts aimed at using AI to improve state services.
Maroney also pointed to a healthcare-innovation partnership with Yale folded into the bill, and set an ambitious marker: making Connecticut “the most AI-literate workforce in America.”
The bigger picture
What ties Public Act 26-15 together is a theory of the role of states. State Representative Hubert Delany (D-Stamford) described it as Connecticut choosing to “lead with clear rules, public trust, and real accountability” rather than waiting. Senate Majority Leader Bob Duff made the same point — that states don’t have to wait for Washington to act for their residents.
That framing is doing real work. Federal proposals on kids and social media, and on AI generally, have repeatedly stalled. By moving at the state level, Connecticut joins a growing patchwork of states writing their own digital rulebooks — which protects residents now, but also raises the familiar question of what happens when fifty states write fifty different sets of rules for companies that operate everywhere.
For now, the bet Connecticut is making is straightforward: clear, enforceable protections in hand are worth more than a perfect federal framework that never arrives. Whether the age-verification and algorithmic-cap provisions survive the inevitable legal and practical challenges — age verification in particular has been a flashpoint in other states — will be the real test of how much this law changes.
Leave a Reply