AI Hiring Tools and Colorado Law: What Businesses Must Do Before January 1, 2027

If your business uses AI anywhere in hiring — a résumé screener, an applicant-ranking feature inside your ATS, a chatbot that pre-screens candidates, video-interview scoring — you are standing in the exact spot Colorado's new AI law was written for. Employment is the covered category small businesses trip over most, usually through software features they never consciously chose. Here is what is legal, what changes on January 1, 2027, and the playbook to run before then.

First: yes, using AI in hiring is legal

Nothing in Colorado law bans AI hiring tools. What the law regulates is how you use them: whether candidates know AI is involved, whether a human meaningfully reviews adverse outcomes, whether you keep records — and, under anti-discrimination laws that have been on the books for decades, whether the tool produces biased results. The tool is fine. The obligations attach to the use.

Do you even have an AI hiring tool? (Check before you answer)

Most owners answer “no” and are wrong. AI hiring features ship embedded in mainstream software, on by default:

• Applicant tracking systemsthat score, rank, or “match” candidates to the job description.
• Job-board features that decide which applicants you see first — or at all.
• Screening chatbots that ask knockout questions and filter people out before a human looks.
• Video-interview toolsthat score communication or “fit.”
• Background-check and assessment platforms with algorithmic scoring layers.

If software influences who gets hired, promoted, disciplined, or fired at your company, it belongs on your AI inventory — step one of our AI legal checklist.

What the Colorado AI Act requires from January 1, 2027

SB 26-189 applies when automated decision-making technology materially influences a consequential decision — and employment decisions (hiring, firing, promotion, discipline, compensation-affecting calls) are squarely on the list. For decisions made on or after January 1, 2027, a covered employer-deployer owes:

1. Pre-use notice. A clear and conspicuous notice — before the AI is used — that AI will play a role in the decision. In practice: language in your job postings, application flow, or careers page.
2. A 30-day adverse-outcome notice. When covered AI contributes to rejecting, passing over, or otherwise deciding against a candidate or employee, they get a notice within 30 days.
3. Meaningful human review.A real person, with real authority to change the outcome, reviewing adverse decisions where commercially reasonable — not a rubber stamp on the model's output.
4. Data access and correction. A way for the person to see and correct the personal data the system relied on.
5. Records for three years. Documentation of the above — which is easier to build as a habit now than to reconstruct under a 60-day cure letter later. Our AI decision-documentation guide shows the system we use with clients.

Whether a given tool “materially influences” your decisions is the threshold question — and it is being defined in the Attorney General's rulemaking right now. Comments are open through July 13, and employers with real fact patterns should be in that file.

The vendor conversation to have this quarter

Most small businesses do not build hiring AI; they license it. That makes your vendor contract your main compliance instrument. Before renewal, ask your ATS or screening vendor:

• Which features in our plan use AI to score, rank, or filter candidates — and can we configure or disable them?
• What documentation will you give us to support our pre-use and adverse-outcome notices?
• Has the tool been tested for adverse impact, and will you share the results?
• Who is liable if the tool produces discriminatory outcomes?

If the contract is silent on these, these are the five clauses to add.

Frequently asked questions

Is it legal to use AI to screen résumés in Colorado?

Yes. But starting January 1, 2027, if the AI materially influences hiring decisions, the Colorado AI Act requires a pre-use notice to candidates, a 30-day notice after adverse outcomes, meaningful human review, data access and correction, and three years of records. And anti-discrimination law applies to AI-driven screening today.

Does the Colorado AI Act apply to my applicant tracking system?

It depends on whether the ATS's AI features materially influence your hiring decisions — a threshold the Attorney General is defining in rulemaking due by January 1, 2027. A scoring feature a human always overrides may land differently than an auto-reject filter. When in doubt, inventory the feature and treat it as potentially covered.

Do I have to tell candidates I use AI in hiring?

From January 1, 2027, yes — covered deployers owe a clear and conspicuous pre-use notice before the AI is used, plus a notice within 30 days when AI contributes to an adverse decision. Adding plain AI-disclosure language to your application flow now is cheap and candidate-friendly.

Can my business be sued if an AI hiring tool is biased?

The Colorado AI Act itself has no private right of action — only the Attorney General enforces it. But a biased hiring outcome can support ordinary discrimination claims under Title VII and the Colorado Anti-Discrimination Act, which individuals can and do bring. The AI act preserves that liability and voids indemnity for your own discriminatory conduct.

The bottom line

AI hiring tools are legal, useful, and — as of January 1, 2027 — regulated. The employers who will have a painless 2027 are the ones who spend 2026 doing four cheap things: inventory the tools, fix the vendor contract, add the notices, and put a human with real authority in the loop.

Not sure whether you are covered? The free AI Act readiness checker takes a few minutes. Want it handled? Every Available Law plan includes attorney review of your hiring-tool setup as an attorney task.