The most common question I get about SB 26-189 is also the simplest: does this even apply to us? The good news is that you can usually answer it in a couple of minutes.
The law covers automated decision-making technology (ADMT) that helps make a consequential decision about a person. “Consequential” isn’t vague hand-waving; the law names seven areas.
The seven kinds of decisions
Ask yourself if AI helps you decide anything in these:
Employment — screening résumés, ranking candidates, scheduling, evaluations.
Financial services — loan or credit decisions, risk scoring, account approvals.
Housing — tenant screening, eligibility, waitlist prioritization.
Insurance — underwriting, pricing, claims triage.
Healthcare — triage, scheduling by acuity, coverage or care decisions.
Education — admissions, financial aid, placement.
Government services — benefits eligibility, program access.
If AI materially influences a decision in any of these, even as a recommendation that a human signs off on you’re in the neighborhood of the law.
Two roles: developer or deployer
The law splits responsibilities between two roles:
A deployer uses AI to help make or inform a decision. Most nonprofits, credit unions, clinics, and agencies are deployers.
A developer builds or substantially modifies the AI. If you buy a tool off the shelf, your vendor is usually the developer. If you meaningfully customize or train it, you may be both.
Most small organizations are deployers which is the lighter side of the obligations.
The three-question test
Do you operate in Colorado or make decisions about Colorado residents?
Does AI help make or inform a decision about a person in one of those seven areas?
Could that decision have a real effect on them in regards to a job, a loan, a home, their care?
Three yeses means you should be getting ready. Not because the sky is falling, but because the deadline is real and the work is manageable if you start now.
If you’re not sure how to answer #2, that’s usually a sign the AI in your organization has spread further than anyone’s tracking. This is known as the “shadow AI”. An AI that a staffer started using without telling anyone. The scorecard is built to help you surface all of that.
→ Take the free AI Readiness Scorecard at GOVERNBOX.ai — five minutes, real answers.
General information about SB 26-189, not legal advice. — GOVERNBOX.ai