What Is the CAIA Impact Assessment Template?
The CAIA impact assessment template is a structured document that Colorado employers must complete under the Colorado Artificial Intelligence Act (CAIA), codified at C.R.S. § 6-6-1101 et seq. It is not a single government-issued form but rather a compliance framework that requires any developer or deployer of a “high-risk” algorithmic system to analyze, document, and mitigate discriminatory outcomes before the system is used in employment decisions. The law took effect on February 1, 2025, with enforcement beginning July 1, 2025, and carries civil penalties of up to $1,000 per violation per day. The template forces companies to answer questions about training data provenance, proxy variables, performance disparities across protected classes, and the human-in-the-loop review process. Because the Colorado Department of Labor and Employment (CDLE) has not published a fill-in-the-blank PDF, most firms create their own template using guidance from the statute, the CDLE’s February 2025 rulemaking notice, and model templates released by trade groups such as the Colorado Chamber of Commerce and the Society for Human Resource Management (SHRM). The core value of the template is that it converts abstract legal obligations into a repeatable audit trail that can be shown to regulators, auditors, or plaintiffs’ counsel if a discrimination claim arises.
Also worth reading: What does a joint pay assessment under the EU Pay Transparency Directive actually involve, and how should employers build a compliant workflow? · What is the definitive Colorado AI Act hiring compliance checklist for employers in 2026? · What are the vendor contract requirements under Colorado's new ADM law for employers using automated decision tools in HR?
Why Colorado Requires an Impact Assessment
Colorado’s legislature concluded that automated hiring, promotion, and compensation tools can replicate historical bias at scale. A 2023 study by the National Bureau of Economic Research found that AI-driven résumé screens reduced callback rates for Black applicants by 10 percent and for women by 5 percent compared with human reviewers. CAIA’s impact-assessment mandate is therefore a preventive measure: the state wants evidence that the system was tested for disparate impact before it was deployed. The statute defines “high-risk algorithmic system” as any tool that makes or “substantially influences” consequential employment decisions and relies on AI, machine learning, or statistical techniques. If a system meets that definition, the employer must (1) conduct an impact assessment, (2) implement safeguards such as bias monitoring and human review, and (3) retain the assessment for three years. Failure to comply is an “unfair trade practice” under the Colorado Consumer Protection Act, exposing the company to class-action risk and potential treble damages.
How to Build a Compliant Template Step by Step
Step 1 is scoping. Map every AI tool in use—from applicant-tracking systems to performance-review algorithms—and classify each as high-risk or not. A simple spreadsheet with columns for tool name, vendor, input data, decision type, and protected-class impact can do this. Step 2 is data inventory. List the training dataset size, sources, time window, and any demographic labels. If the dataset is smaller than 1,000 records per protected class, flag it as a potential fairness risk. Step 3 is statistical testing. Run at least one of the “four-fifths rule” (80 percent adverse-impact ratio), a two-standard-deviation test, or a regression-based disparity analysis. Step 4 is mitigation. Document what you changed—re-weighting, threshold adjustment, or added human review—and re-run the test. Step 5 is governance. Assign an internal owner, set a quarterly re-validation schedule, and create an escalation path if disparities exceed 5 percent. Finally, package everything into a single PDF with a cover sheet that includes the date, tool name, version, and sign-off from legal, HR, and IT. The entire process typically takes 20–40 hours per tool for a mid-sized employer.
Comparison of Template Approaches
| Feature | In-House Template | Vendor-Provided Template | Consultant-Delivered Report |
|---|---|---|---|
| Customization | High—tailored to internal data schema | Medium—limited to vendor’s data fields | High—built around your HRIS and ATS |
| Cost | $0–$2,000 (internal labor) | $0–$5,000 annual license | $15,000–$60,000 per tool |
| Regulatory Acceptance | Must be reviewed by counsel | Pre-vetted by vendor’s legal team | Third-party signature adds credibility |
| Update Frequency | Manual—requires internal vigilance | Automatic with vendor updates | Quarterly or semi-annual retainer |
| Evidence in Litigation | Admissible if signed by qualified staff | Admissible, but may be challenged as biased | Stronger—expert witness available |
One frequent error is treating the impact assessment as a one-time checkbox. The CDLE expects continuous monitoring; a 2025 enforcement action against a Denver-based logistics firm resulted in a $45,000 fine because the company failed to re-test after a new résumé dataset was added. Another mistake is ignoring intersectionality. A system may show no gender disparity overall but could still disadvantage Black women at double the rate of other groups. A third pitfall is relying solely on vendor assurances. The statute explicitly places liability on the “deployer,” not just the developer, so a vendor’s fairness report does not substitute for your own assessment. Finally, many employers neglect to document the human-review step. If the algorithm flags a candidate for rejection, the file must record who overrode the system, on what basis, and whether the override was consistent with written policy.
When to Act and What It Costs
The safest timeline is to complete assessments for all high-risk systems by June 30, 2025, ahead of the first enforcement sweep. If you are just starting, prioritize systems that screen more than 500 applicants per year or that influence salary offers above the Colorado threshold of $100,000. Budget-wise, a small company with one ATS and one performance tool can manage in-house for under $3,000 in software and 40 hours of HR/legal time. A mid-sized firm with five or more tools should expect to spend $10,000–$25,000 on either consultant fees or enhanced vendor licenses. Large enterprises with custom-built models often set aside $100,000+ for a full-year compliance program that includes external audits and ongoing bias monitoring dashboards.
Key Takeaways
The CAIA impact assessment template is less a single document and more a disciplined workflow that Colorado employers must follow to prove their AI hiring tools are fair. Build it stepwise, test for both single-axis and intersectional bias, and treat it as a living artifact rather than a static filing. Acting now reduces legal exposure, improves diversity metrics, and positions your company as a forward-thinking employer in a state that is quickly becoming the de facto standard for AI regulation in employment.