By Sean Camacho

Consumers in Colorado and across the country are demanding artificial intelligence companies be more transparent about how their systems work: how they’re built, how they learn, how they change over time. What we’ve learned as advocates for AI regulation is the process of building AI — iterative, adaptive and never fully finished — might actually have something to teach us about how to write the laws that govern it.

Twenty years into the social media era, the cost of waiting to address the clearly emerging dangers of rapidly advancing technology is overwhelming. Though platforms reshaped how we communicate, debate and understand the world, policymakers largely held back. Consumers eventually took matters into their own hands through legal action, but by the time they did, the harm had already been baked into the structure of ubiquitous platforms.

With AI, we have a chance to learn from those mistakes. But our window to act won’t stay open indefinitely. The risks aren’t hypothetical. We’ve already seen what happens when conversational AI operates without guardrails: real-world consequences for Coloradans of all ages, particularly our youth. Every legislative session we spend waiting for perfect answers is another year in which those consequences continue without any framework to address them.

But acting now doesn’t mean acting carelessly. It means being willing to legislate the way this technology actually develops: constantly, consistently and methodically, with the humility to treat the first version of a law as a beginning rather than a conclusion.

Part of what makes AI difficult to regulate is the sheer pace at which it continues to develop. What we’ve learned during the 2026 legislative session is in a fast-moving world, good policy can’t stand still. When the underlying technology is evolving in real time, policymaking doesn’t unfold in a straight line. It’s iterative. Assumptions change. New information forces reconsideration. One small change can make the difference between full-throated support or full-throttle opposition from the same stakeholder group.

This session is the first in Colorado where AI is truly ingrained in the everyday lives of nearly all of our constituents. Some voices of opposition have mistaken the complexity of developing good policy for inconsistency and suggested we should wait for clear answers before acting. But waiting has consequences, too. What we should demand of one another is a willingness to treat legislation the way this technology treats itself: always changing, always improving, open to revision and collaboration from all sides as the landscape evolves. That means we should never again treat a bill as one-and-done.

Stakeholder engagement is typically a phase: you gather input, negotiate changes and move forward. But legislating in a way that keeps pace with the world means engagement has to be more constant and transparent — and it means sitting with the core tension this kind of work creates: move too quickly and you risk getting it wrong; move too slowly and you allow real harm to take root without any guardrails at all. There is a meaningful difference between a process that ignores that tension and one that engages with it directly.

That approach won’t produce legislation that satisfies every perspective. In areas like this, the goal isn’t to arrive at a perfect, final answer — it’s to take a step that reflects what we know now, while leaving room to adjust as we learn more. That’s what HB26-1263 represents: months of engagement, iteration and compromise designed to establish clear guardrails around conversational AI while leaving room to adapt as the landscape continues to change. Some will argue it should go further. Others say that it goes too far or moves too soon. We’ve taken both seriously and worked diligently to thread the needle — building a bill with the best possible chance of passing with bipartisan support, earning the governor’s signature, and taking a meaningful step toward protecting Coloradans, especially youth.

That may not be the kind of certainty people are used to seeing from legislation. But it may be a more honest version of what transparency actually requires: not just clarity in outcomes, but openness about the process, the tradeoffs and the fact not everything is settled on day one.

AI isn’t going anywhere. Neither is our obligation to govern it responsibly. We started by demanding transparency from the systems shaping our lives, but transparency isn’t just something we should require of technology. It’s something we should require of ourselves. That means writing laws that are honest about their limits, open to revision and built with the understanding the work doesn’t end at passage. That’s not a concession. That’s the point.

Sean Camacho represents Denver in District 6. He serves as vice chair of the Business Affairs and Labor Committee. In addition to his work in the legislature, and after leaving active-duty Air Force service, he works as an attorney.