Home Part of States Newsroom
Commentary
I helped write Colorado’s first AI chatbot bill. Here’s what I learned about state action.

Share

I helped write Colorado’s first AI chatbot bill. Here’s what I learned about state action.

May 05, 2026 | 1:18 pm ET
By Alexis Alltop
I helped write Colorado’s first AI chatbot bill. Here’s what I learned about state action.
Description
(David Espejo/Stock photo via Getty Images)

For nearly a year, I’ve spent my days on calls with colleagues across the country, researchers, advocates and policy experts all grappling with the same reality: rapidly developing conversational AI technology that we know is doing harm to our communities, an absence of guidelines, and a combination of political and legal challenges that can make legislation with teeth hard to pass.

I’ve had a front row seat to the work happening in the states on the front lines of the effort to learn the mistakes of the social media era that are now coming home to roost. And I can tell you with confidence, what’s coalescing isn’t a patchwork of regulations, it’s a foundation and a framework that Colorado is learning from and building upon.

The bill that was crafted as a result of those conversations and in partnership with our Colorado legislators, House Bill 26-1263, has now passed the Colorado House and been introduced in the Senate. It would require AI chatbot operators to disclose that users are talking to a machine, follow evidence-based protocols when a user expresses suicidal thoughts or self-harm, and meet stronger standards when their users are minors. It is not a panacea, but rather a critical step forward that Coloradans of all ages, but especially young people, deserve.

According to Pew Research, nearly two-thirds of American teenagers report using AI chatbots, and about 1 in 3 use them daily. A nationally representative study found that roughly 5.4 million young Americans have turned to generative AI for mental health advice, and that more than 40 percent of them were never directed toward professional help or crisis resources. In 2025 alone, AI-induced psychosis was linked to multiple suicides and at least one murder.

Colorado kids are the early adopters of this technology, and the risks associated with that adoption are anything by hypothetical. And whatever harms we are tracking today are, almost certainly, the leading edge of something larger.

Need to get in touch?

Have a news tip?

I came into this work assuming the state-by-state policy landscape would look chaotic — a patchwork. That’s the word some opponents are using, and to be honest, it’s the word I was bracing for. What I found instead was that California, New York, Oregon, and Washington had each landed on remarkably similar requirements: clear disclosure, protocols for crisis prompts, layered protections for minors. Hawaii and Illinois are moving bills that mirror those same standards. Colorado’s bill was drafted with explicit attention to those frameworks with Colorado-specific adaptations where they were needed.

Some have leveled accusations that this is an “industry bill.” As the person who has been doing the research and work to write the policy behind this legislation, I can tell you that is unequivocally false. The fact that the bill takes frameworks that have passed in other states and builds a Colorado foundation around them isn’t an indication of nefarious intent. It’s an encouraging sign that communities across the country are working to find ways to quickly and effectively meet the moment as AI technology continues to develop so quickly.

It’s also a way to balance Colorado’s needs while managing the “patchwork” concern, which is legitimate. A genuinely fragmented set of laws would create real burdens for compliant companies and real gaps for users who cross state lines. HB-1263 strikes the right balance.

What states are doing right now is about more than filling a gap that Congress will get around to in a few years. It is laying the groundwork for whatever federal framework eventually arrives. And it is doing so along lines that look, from where I’m sitting, more coherent than chaotic.

The bill is part of the movement to design a better future for the next generation; one in which we can embrace the development of AI technology for the potential good it might do, without worrying that every interaction puts users at risk for serious harm.

That, I think, is the story worth telling about HB-1263,  and about every state law like it, working its way through legislatures like ours.

Related News

You shouldn’t need a law degree to hunt moose in Alaska
Commentary
Help Montana’s hungry bears stay safe
Commentary
Minnesotans: We can trust our neighbors who run our elections
Commentary