Skip to content
ContestabilityA research
publication
HomeRulemaking comment

Making review rights usable: Colorado ADMT rulemaking

A filed comment proposing decision-specific reasons, usable reconsideration, and records of what human reviewers actually considered.

Author
Kerry Thornhill
Status
Rulemaking comment · Filed October 24, 2026
Source date
Web version
2026-11-15 · v1

Kerry Thornhill

MS in Interdisciplinary Studies · In progress · University of North Texas

Anuradha and Vikas Sinha College of Artificial Intelligence and Advanced Analytics

The Colorado State Capitol and the Denver skyline below the Rocky Mountains, seen from a mainframe computer room with a ribbon of printout

Rulemaking comment · Filed October 2026

Making review rights usable: Colorado ADMT rulemaking

Kerry Thornhill

A right to human review needs more than an invitation to appeal. An affected person must be able to identify the grounds of a decision, investigate what may be wrong, and bring a material objection before someone who can change the outcome.

On October 24, 2026, I filed Usable Disclosure and Review Rights in the Colorado ADMT Rules with the Colorado Department of Law. The comment addresses the October 6 interim draft of the rules implementing Colorado’s Automated Decision-Making Technology and Chatbot Safety Act. It applies the Necessary Informational Conditions (NIC) framework to the practical design of disclosure, correction, and reconsideration.

From categories of data to grounds for challenge

Knowing that an employer considered “employment information” does not tell an applicant what to investigate. If the actual reason was four recorded months of relevant experience against a twelve-month criterion, the applicant can question the record, the criterion’s application, or the inference drawn from it. Naming a data category and explaining its adverse significance perform different functions.

The comment asks for decision-specific reasons that identify how the automated output contributed to the adverse outcome. It distinguishes the factors that generated a system’s output from the institution’s reasons for acting on that output.

Making reconsideration usable

The submission proposes safeguards for the whole course of a challenge: timely access to grounds and supporting information; protection against avoidable harm while review is pending; reviewers with authority to change the decision; and contemporaneous records showing what a human decision-maker actually considered. It also distinguishes correction of inaccurate personal data from questioning reliance on an inference or score.

These are proposed improvements to the draft rules. The comment identifies the statutory provisions on which it relies and marks additional procedural safeguards as proposals. It does not claim that Colorado already requires every element of the NIC framework or that the Department has adopted these recommendations.

Download the filed Colorado rulemaking comment (PDF, 8 pages)

The conceptual foundation is developed in From Notice to Contestable Grounds: Necessary Informational Conditions for Algorithmic Contestability. The later ICO response on agentic AI extends the inquiry to automated workflows and meaningful human involvement.

Colorado Department of Law: rulemaking materials and official updates

Return to Articles and Activities