Skip to main content

Colorado ADMT Comment Letter Template

Copy-paste comment letter and 3-minute hearing script for the proposed Colorado ADMT and chatbot rules. Comments close Oct 26, 2026 at 11:59 PM MT.

Part 3 of 4The story so far: Colorado AI Act: from SB 26-189 to the 2027 rules ↓

11 min readBy Jesus A McKinney
Colorado ADMT Comment Letter Template

TL;DR Colorado's Attorney General is taking written comments on proposed rules for the ADMT Act (SB 26-189) and the chatbot law (HB 26-1263) until 11:59 PM on October 26, 2026. The hearing is the same day at 10 a.m. in Denver, with a remote option. This page has a fill-in comment letter, a menu of positions on the open issues, a numbers worksheet and a 3-minute hearing script. Comments are posted publicly, so write for a regulator and a competitor at the same time.

The proposed rules were filed August 11, 2026, and SB 26-189 takes effect January 1, 2027. The Attorney General is due to adopt rules by then. Comments received by the October 26 deadline are the last chance to change the text before the final rules.

Most businesses will not comment. That is the opportunity: a short, specific, well-evidenced letter from a real deployer carries weight that a general objection does not. This page gives you the structure. You supply the facts.

This is a drafting aid, not legal advice. Have counsel read your letter before you file it, especially if you operate in several states.

Before you write: facts to confirm

The sources we read disagree on a few details, and a revised draft may have changed the numbering. Check each of these on the Attorney General's rulemaking page at coag.gov/ai before you file.

Item What we found Confirm
Written comment deadline 11:59 PM, October 26, 2026 Portal wording
Hearing October 26, 10:00 AM, Ralph L. Carr Judicial Center, 1300 Broadway, Denver, with a remote option Remote link
Where to file The comment portal linked from coag.gov/ai (one summary gives comments.coag.gov) Current URL
Earlier comment cut-off Sources say September 4 or October 5 for comments considered in a revised draft Moot after those dates
Revised draft The Attorney General said revised rules were expected in late September Whether it was posted, and new rule numbers
Public posting Reports say all comments are posted publicly Portal notice

The rule numbers below follow the August 11 proposal as summarized by law firms: Rule 5 (developer disclosures), Rule 6 (adverse outcome notices, with the data-source duty reported as Rule 6.6 in some summaries), Rule 7.7 (human review) and Rules 8 to 13 (chatbots). If a revised draft renumbered them, cite the new numbers.

What the proposal would require

You need this in front of you to write specific comments. As reported by the law firms we read:

  • Adverse outcome notice. Within 30 days, in plain language through at least two channels and in at least 12-point type if printed. It must name the decision, describe the tool's role and any human reviewer's role, state the principal reasons for the outcome "with specificity", explain inferences and the data behind them, disclose actual scores where a score drove the outcome, flag missing information, and identify data sources by name. One summary adds that a deployer that cannot explain how the technology influenced the decision does not comply.
  • Human review. A consumer can ask for meaningful human review. The deployer must confirm receipt within 10 days and finish within 45. The reviewer must be independent of the original decision maker and not that person's subordinate wherever feasible, have documented training, and have authority to overturn the decision. Summaries add a ban on AI assistance in the review, a 24-hour response window for phone or mail requests, and delay of the adverse outcome pending review where possible. Deployers must document the reviewer's identity, authority and training, timestamps, evidence reviewed and written justification.
  • Commercially reasonable. Human review applies "to the extent commercially reasonable", with a presumption of reasonableness for outcomes such as termination or denial of housing.
  • Data access and correction. Separate rights to see the data used and correct inaccuracies.
  • Developer information. Reported to include intended uses, known limitations, known inappropriate uses, training data categories and monitoring instructions.
  • Chatbots. Age estimation that does not rely only on a user's claim, a persistent visible disclaimer for minors no smaller than the largest text on screen, self-harm protocols, no presenting outputs as licensed professional services, and an annual report to the Attorney General.

One summary reports no small-business exemption in the statute or the rules.

The issues the Attorney General left open

These are the places where a comment can change the outcome, because the Department itself flagged them.

  1. What "materially influence" means. Two options are on the table. One excludes only factors that are "trifling, trivial, or incidental". The other allows an exemption where the human's independent analysis plays a significantly larger role than the AI output. One firm called the gap between them the most consequential open issue.
  2. Who is responsible when a vendor runs the tool. The Department acknowledged the act does not explicitly address deployers that do not directly operate the technology, and asked whether vendors should carry obligations or the principals remain responsible. Staffing agencies and recruitment outsourcers raise the same question.
  3. How much developers must disclose. What detail on testing, training data categories and format.
  4. What counts as a "basic human need". Housing denial is treated as presumptively requiring review. Other categories, such as utilities, benefits and health care, are not confirmed.

A laptop on a desk showing code on a dark screen, representing drafting a regulatory comment with supporting data

Numbers worksheet (fill this in first)

A comment with numbers is read. A comment without them is skimmed. Collect these before you write.

Our organization
- Name, headcount, industry: ______
- Role: deployer / developer / both: ______
- Tools in scope (category, vendor, decision type): ______

Volume
- Consequential decisions per year affecting Colorado residents: ______
- Of those, adverse outcomes per year: ______
- Expected human review requests per year (our estimate and basis): ______

Cost
- Hours per human review under the proposed standard: ______
- Staff needed (name roles) and annual cost: ______
- Cost to rebuild notices (12-point, two channels, scores, sources): ______
- Cost or delay if our vendor cannot supply principal reasons: ______

Effect
- Hiring or service delay from a 45-day review window: ______
- Tools we would stop using if the rule is adopted as written: ______

If you cannot estimate a number, say so in the letter and say why. An honest "we cannot yet estimate this because our vendor has not told us what data it uses" is itself useful information for the Department.

The comment letter template

Copy this into a document, fill the brackets, delete the options you do not use, and keep each issue to a few short paragraphs.

[Date]

Colorado Department of Law, Office of the Attorney General
Re: Comments on proposed rules implementing SB 26-189 (Automated
Decision-Making Technology Act) and HB 26-1263 (Chatbot Safety Act),
filed August 11, 2026

To the Department:

1. WHO WE ARE
[Company] is a [industry] business with [N] employees, [N] of whom are
in Colorado. We use [tool category] from [a third-party vendor / our own
system] to support [decision type, for example screening applicants].
We are a [deployer / developer / both]. In 2026 we made about [N]
decisions of this kind affecting Colorado residents.

2. SUMMARY OF OUR REQUESTS
- [Request 1 in one sentence, with the rule number]
- [Request 2]
- [Request 3]

3. ISSUE A: "MATERIALLY INFLUENCE" (Rule [__])
Our situation: [How a human and the tool interact in our process.]
Our position: [We support the option that excludes only trifling,
trivial or incidental factors, because ___.] OR [We support the option
that exempts decisions where independent human analysis plays a
significantly larger role, because ___.] OR [We propose ___.]
Requested change: [Specific wording or threshold.]

4. ISSUE B: VENDOR AND STAFFING RELATIONSHIPS (Rule [__])
Our situation: [We do not operate the tool. Our vendor ___.]
Our position: [Obligation X should sit with the developer because only
the developer can produce ___. We remain responsible for ___.]
Requested change: [Allocate the duty by who controls the information,
require developers to supply ___ to deployers within ___ days.]

5. ISSUE C: ADVERSE OUTCOME NOTICES (Rule 6 / [__])
Our situation: [Our current notice process and how it relates to
existing adverse action notices.]
Our position: [We support specific principal reasons. We ask the
Department to clarify ___ because ___.]
Evidence: [Cost to rebuild notices: $___. Time: ___.]
Requested change: [For example, a safe harbor if the notice uses
developer-supplied reasons, or a template notice.]

6. ISSUE D: HUMAN REVIEW (Rule 7.7 / [__])
Our situation: [Headcount, who could serve as an independent reviewer
in a team of our size.]
Our position: [The "not a subordinate, wherever feasible" standard is
workable / unworkable for us because ___. A 45-day period would
[delay hiring by ___ / be workable].]
Evidence: [Estimated reviews per year: ___. Hours per review: ___.
Annual cost: $___.]
Requested change: [For example, define "wherever feasible" for
organizations under [N] employees, allow review by an external reviewer,
or allow a shorter period for low-risk decisions.]

7. ISSUE E: "COMMERCIALLY REASONABLE" AND SIZE
Our situation: [We are a small organization. Neither the statute nor
the proposed rules include an exemption.]
Requested change: [Add factors to the commercially reasonable test:
organization size, volume of decisions, and cost per review.]

8. ISSUE F: CHATBOTS (Rules 8 to 13) [include only if you operate one]
Our situation: [Our chatbot is used for ___ and is not directed at
minors.]
Our position: [Age estimation ___; the persistent disclaimer ___.]
Requested change: [Clarify whether ___ applies to ___.]

9. OTHER POINTS
[Any practical problem not covered above. Cite the rule number.]

10. CLOSING
Thank you for considering these comments. We are available to discuss
them and can provide further information on request.

[Name, title, organization, contact]

Rules for writing it well

  1. One issue per heading, and cite the rule number. The reader is sorting hundreds of comments by section.
  2. Say what you want changed, in words the Department could adopt. "This is burdensome" gets noted. "Define 'wherever feasible' as follows" can be used.
  3. Use your numbers. The worksheet exists for this.
  4. Support what you support. A comment that also endorses the parts you accept is more credible than a list of objections.
  5. Do not claim what you cannot back up. If you have not read the full text of a rule, say you are relying on a summary.
  6. Keep it short. Two to four pages is plenty.
  7. Assume it is public. Leave out customer data, vendor pricing under NDA and anything confidential. Consider filing through a trade association or coalition so operational details are not tied to a single company.

A 3-minute hearing script

The hearing is October 26 at 10 a.m. in Denver, with a remote option. Check the portal or the Department's notice for how to sign up to speak and for time limits. Read it aloud once before you go.

Good morning. My name is [name] and I am [title] at [company], a
[industry] business with [N] employees, [N] in Colorado.

We use [tool category] to [decision type], roughly [N] times a year for
Colorado residents. We support the goal of telling people why an
automated tool affected a decision about them.

I want to raise [one or two] issues.

First, [Issue, with rule number]. As written, [effect on us, with one
number]. We ask the Department to [specific change].

Second, [Issue]. [One sentence of evidence.] We ask the Department to
[specific change].

We have submitted written comments with the detail. Thank you for the
opportunity to speak, and I am happy to answer questions.

Stay under three minutes by speaking to two issues, not six. Everything else goes in the written letter.

Final checklist before you file

  • Rule numbers checked against the current version on coag.gov/ai
  • Every number in the worksheet has a basis you can state
  • Nothing confidential in the text or attachments
  • Counsel has read it
  • Filed before 11:59 PM on October 26, 2026, with the confirmation saved
  • Hearing sign-up done if you plan to speak

What we could not verify

  • We did not read the proposed rules themselves, only law firm and compliance summaries. The summaries agree on the main points but differ on a few details, including the earlier comment cut-off and the hearing address.
  • We could not confirm whether the revised draft expected in late September was posted, so the rule numbers may have changed.
  • We could not load the Attorney General's comment form page. The portal address comes from secondary sources, so confirm it at coag.gov/ai.
  • The "no small-business exemption" statement comes from one summary.
  • Nothing here predicts what the Department will do with any comment.

Tagscoloradostate-ai-lawtemplateemployment-aipublic-comment

The story so far · Part 3 of 4

Colorado AI Act: from SB 26-189 to the 2027 rules

  1. led to

  2. then

    You are here

    Comment letter template and hearing script for the October 26 deadline

  3. then

    Attorney general releases an interim redline draft of the rules

  4. then

    Next

    Written comments close at 11:59 p.m. MST; formal rulemaking hearing

  5. then

    Next

    SB 26-189 duties take effect

See the full timeline →

Legal disclaimer

This article is published for informational and educational purposes only. It does not constitute legal, regulatory, or professional compliance advice and should not be relied upon as such. AI governance requirements vary by jurisdiction, industry, and organizational context. Always consult a qualified legal or compliance professional before implementing policies or making decisions with regulatory implications.

About the author

Jesus A McKinney

Templates & Workflows Desk

Builds the copy-paste side of the site: policy templates, registers, checklists and vendor comparisons. Each template says which requirement every field answers, and is written to be used the same day.

  • Fields mapped to the clause they answer
  • Vendor terms read on the vendor's own pages
  • Dated checks on each post