Colorado just made a major pivot on AI regulation — and it’s a big win for the tech industry.

Remember the Colorado AI Act? It was the most comprehensive AI law in the United States, mimicking the EU AI Act’s approach by classifying “high-risk” AI systems and imposing obligations on developers and deployers. It was supposed to go into effect in February 2026, got postponed to June, and now — if the new proposal passes — it might not exist at all in its current form.

On March 17, 2026, the Colorado AI Policy Work Group, with strong support from Governor Jared Polis, delivered a unanimous proposal for a completely new framework called “Concerning the Use of Automated Decision Making Technology in Consequent Decisions” (the Proposed ADMT Framework). And it’s a fundamentally different approach.

The shift: Instead of risk management, the new framework focuses on transparency and consumer rights. The old law required AI companies to report algorithmic discrimination to the Colorado Attorney General, conduct AI impact assessments, and implement formal risk management policies (think NIST AI RMF or ISO/IEC 42001). Those requirements? Gone.

What’s left: Developers now need to provide technical documentation to deployers about intended uses, harmful uses, training data, limitations, and risks. Deployers need to tell consumers they’re using automated decision-making technology and provide notice within 30 days if an AI rendered an “adverse outcome” — along with a process for human review.

The new framework also raises the bar for what even counts as regulation. The old law used a “substantial factor” standard — if an AI simply assisted and could alter a high-risk decision, it was covered. The new framework uses “materially influence” — the AI output has to be a non-trivial factor affecting the actual outcome. That’s a higher bar, meaning fewer systems are covered.

Why this matters: Colorado was the only state with a comprehensive AI law of this scope. Now, with this rewrite, it’s signalling a more business-friendly approach — one that emphasizes telling consumers what’s happening rather than preventing harm before it occurs. If this passes, expect other states to watch closely. The question is whether this “light touch” approach protects people or just shifts the burden onto consumers to figure out they’ve been harmed after the fact.

The Proposed ADMT Framework would go into effect January 1, 2027 — giving covered companies until the end of 2026 to adjust.