Colorado Employers Should Prepare Now for Automated Decision Making Rules

Colorado’s automated decision-making law becomes effective January 1, 2027, and employers should treat the remaining runway as implementation time, not waiting time. The Colorado Attorney General completed its formal rulemaking hearing and closed written comments on October 26. Final rules are still pending, but the statute already tells employers where the state is headed: more transparency, more documentation, better data governance, and real human involvement when technology materially influences consequential employment decisions.

The law reaches automated decision-making technology used to materially influence hiring, promotion, compensation, discipline, termination, and similar employment decisions involving Colorado applicants and employees. The coverage is broader than a tool that makes the final call. A resume-ranking platform, personality assessment, video-interview analysis, retention-risk score, productivity flag, scheduling system, or attendance algorithm may be covered if its output materially influences what a human decision-maker does next. In other words, an employer cannot avoid the question by saying a manager still clicks the final approval button.

The core compliance challenge is information. Employers need to understand what each system does, what data it uses, how the output is presented, who relies on it, and what happens when an individual disputes an inaccurate input or adverse result. That information often sits with a vendor, not with HR. Waiting until the final rules arrive to ask those questions is risky because vendor contract changes, product-roadmap conversations, and workflow redesign take time.

The proposed rulemaking materials point toward notice, explanation, correction, and meaningful human-review processes. Employers should expect to need a clear explanation when an automated system materially contributes to an adverse employment outcome. They should also expect to give people a practical route to correct inaccurate personal data and request reconsideration. A human reviewer should have enough independence, training, authority, and underlying information to do more than rubber-stamp a system recommendation.

A strong first step is a workforce-technology inventory. Ask every HR, talent-acquisition, operations, payroll, and information-technology leader to identify tools that score, rank, recommend, predict, filter, flag, or generate employment-related outputs. Then create a short record for each tool: purpose, workforce affected, data inputs, vendor, decision point, human reviewer, and adverse-action consequence. This exercise often reveals embedded features that no one realized were active.

Next, revise vendor agreements and renewal questionnaires. Employers should obtain technical documentation, intended-use descriptions, data-source information, known limitations, validation material, change notices, and cooperation with notices or appeals. Create a human-review procedure before a termination, discipline, or hiring rejection depends materially on a system output. Preserve the information reviewed and the rationale for the final decision.

Colorado is not asking employers to abandon useful technology. It is asking them to understand it well enough to use it responsibly. A thoughtful inventory and decision-review process will help organizations prepare for Colorado while strengthening AI governance across every state where they operate. Employers should assign an accountable owner for each high-impact tool and establish a review cadence whenever the vendor changes the model, data source, scoring criteria, or intended use. That accountability should be visible in the organization’s AI governance register.

Lisa Smith, SPHR, SCP