Colorado Automated Decision-Making Technology Act (SB 26-189)
Signed 14 May 2026, repealing and replacing the 2024 Colorado AI Act with a narrower transparency and disclosure regime for automated decision-making technology that materially influences consequential decisions in education, employment, housing, financial services, insurance, healthcare and essential government services.
| Jurisdiction | Colorado |
|---|---|
| Category | AI Regulations |
| Status | Upcoming |
| Effective date | |
| Latest development |
Analysis
The Colorado Automated Decision-Making Technology Act (SB 26-189) was signed on May 14, 2026 and is the current Colorado AI-law framework replacing the earlier 2024 act; the bill text and legislative page both indicate the law takes effect on January 1, 2027.https://leg.colorado.gov/bills/sb26-189https://leg.colorado.gov/bill_files/116432/download
Because I do not have live web access in this turn, I can only ground the analysis in the official Colorado legislative materials surfaced earlier plus the secondary legal analyses that were retrieved. I’ve still organized the answer in the structure you requested and included clickable source links for each major point.
Key Requirements
- Scope trigger: The law applies to automated decision-making technology (ADMT) that processes personal data and uses computation to generate outputs such as predictions, recommendations, classifications, rankings, or scores used to make, guide, or assist a decision about an individual.https://leg.colorado.gov/bill_files/116432/downloadhttps://leg.colorado.gov/bills/sb26-189
- Material influence threshold: Compliance obligations attach when the ADMT is used to materially influence a consequential decision, rather than to all AI uses.https://leg.colorado.gov/bill_files/116432/downloadhttps://www.hklaw.com/en/insights/publications/2026/05/colorado-governor-signs-sb-189
- Covered domains: The statute covers consequential decisions in education, employment, housing/residential real estate, financial services, insurance, healthcare, and essential government services/public benefits.https://leg.colorado.gov/bill_files/115717/downloadhttps://www.hklaw.com/en/insights/publications/2026/05/colorado-governor-signs-sb-189
- Developer technical documentation: Developers of covered ADMT must provide deployers with technical documentation describing the system’s intended uses, categories of training data, known limitations, and instructions for appropriate use and human review.https://leg.colorado.gov/bill_files/116432/downloadhttps://leg.colorado.gov/bill_files/115880/download
- Deployer notice obligations: Secondary analyses of the enacted bill indicate deployers must give individuals clear notice before using covered ADMT in consequential decisions.https://www.littler.com/news-analysis/asap/colorado-amends-its-artificial-intelligence-law-substantially-reducinghttps://www.ebglaw.com/workforce-bulletin/inside-colorados-senate-bill-26-189-impacts-and-implications-for-employers
- Adverse action / correction / human review process: The law is described in practitioner commentary as requiring a structured process that includes notice, a right to correct, and a right to meaningful human review when ADMT affects a consequential decision.https://www.littler.com/news-analysis/asap/colorado-amends-its-artificial-intelligence-law-substantially-reducinghttps://www.ebglaw.com/workforce-bulletin/inside-colorados-senate-bill-26-189-impacts-and-implications-for-employers
- Record retention: Secondary commentary on the enacted law states employers/deployers should retain relevant records for at least three years.https://www.littler.com/news-analysis/asap/colorado-amends-its-artificial-intelligence-law-substantially-reducing
- Exclusions: The bill materials indicate some technologies are not considered ADMT, including certain cybersecurity-related tools.https://leg.colorado.gov/bill_files/115717/downloadhttps://leg.colorado.gov/bill_files/115880/download
- Effective date: The revised framework takes effect on January 1, 2027.https://leg.colorado.gov/bills/sb26-189https://leg.colorado.gov/bill_files/116432/download
Compliance Challenges
- Identifying whether a tool is “covered ADMT”: Organizations must determine whether a model or vendor system actually processes personal data and whether its output materially influences a consequential decision, which can be difficult when AI is embedded in broader workflows.https://leg.colorado.gov/bill_files/116432/downloadhttps://www.reedsmith.com/our-insights/blogs/technology-law-dispatch/102mrzf/sb-26-189-colorado-legislature-kicks-off-caia-rewrite-race/
- Vendor opacity and documentation gaps: Deployers may rely on third-party tools without receiving enough information on training data, limitations, or intended use to satisfy disclosure and human-review requirements.https://leg.colorado.gov/bill_files/116432/downloadhttps://www.ebglaw.com/workforce-bulletin/inside-colorados-senate-bill-26-189-impacts-and-implications-for-employers
- Workflow redesign in high-stakes decisions: Employers, lenders, insurers, schools, and government agencies may need to redesign decision workflows so that human review is meaningful rather than nominal.https://www.littler.com/news-analysis/asap/colorado-amends-its-artificial-intelligence-law-substantially-reducinghttps://www.hklaw.com/en/insights/publications/2026/05/colorado-governor-signs-sb-189
- Data and records management burden: The three-year retention expectation described in practitioner commentary creates operational pressure on logging, versioning, and evidence preservation.https://www.littler.com/news-analysis/asap/colorado-amends-its-artificial-intelligence-law-substantially-reducinghttps://www.ebglaw.com/workforce-bulletin/inside-colorados-senate-bill-26-189-impacts-and-implications-for-employers
- Real-world example: hiring and workforce tools: Legal analyses note the biggest immediate effects are on employment-related decisions, especially hiring, compensation, and workforce management systems that rank or screen candidates.https://www.ebglaw.com/workforce-bulletin/inside-colorados-senate-bill-26-189-impacts-and-implications-for-employershttps://www.buchalter.com/insights/colorado-rewrites-its-ai-law-what-employers-must-know-about-sb-26-189/
- Real-world example: housing and credit-like decisions: Because the law covers housing and financial services, any model used for screening, scoring, or ranking applicants can trigger disclosure and review obligations.https://leg.colorado.gov/bill_files/115717/downloadhttps://www.hklaw.com/en/insights/publications/2026/05/colorado-governor-signs-sb-189
Implementation Best Practices
- Build an ADMT inventory: Catalog every system that uses personal data and produces predictions, rankings, scores, classifications, or recommendations for consequential decisions.https://leg.colorado.gov/bill_files/116432/downloadhttps://www.reedsmith.com/our-insights/blogs/technology-law-dispatch/102mrzf/sb-26-189-colorado-legislature-kicks-off-caia-rewrite-race/
- Map decision flows end-to-end: Document where the model is used, who reviews outputs, what factors influence the final decision, and whether the model is a non-de minimis factor in the outcome.https://www.reedsmith.com/our-insights/blogs/technology-law-dispatch/102mrzf/sb-26-189-colorado-legislature-kicks-off-caia-rewrite-race/https://leg.colorado.gov/bill_files/116432/download
- Contract for documentation rights: Require vendors to provide the technical documentation the statute contemplates, including intended uses, training-data categories, limitations, and human-review instructions.https://leg.colorado.gov/bill_files/116432/downloadhttps://leg.colorado.gov/bill_files/115880/download
- Design notices and appeal paths: Prepare pre-use notices, post-decision explanations where applicable, and a workflow for correction requests and meaningful human review.https://www.littler.com/news-analysis/asap/colorado-amends-its-artificial-intelligence-law-substantially-reducinghttps://www.ebglaw.com/workforce-bulletin/inside-colorados-senate-bill-26-189-impacts-and-implications-for-employers
- Maintain audit-ready logs: Keep model versions, prompts or input schemas, outputs, reviewer actions, and decision rationales to support retention and defensibility.https://www.littler.com/news-analysis/asap/colorado-amends-its-artificial-intelligence-law-substantially-reducinghttps://www.ebglaw.com/workforce-bulletin/inside-colorados-senate-bill-26-189-impacts-and-implications-for-employers
- Use AI governance frameworks: The NIST AI Risk Management Framework is a practical implementation baseline for mapping, measuring, managing, and governing AI risk, and the NIST Generative AI Profile can supplement controls for AI-enabled workflows.https://www.nist.gov/itl/ai-risk-management-frameworkhttps://www.nist.gov/itl/ai-risk-management-framework/generative-artificial-intelligence-profile
- Use privacy/PIA tools: If personal data is involved, align with the NIST Privacy Framework and privacy impact assessment practices to document data minimization, fairness, and accountability.https://www.nist.gov/privacy-frameworkhttps://www.nist.gov/itl/ai-risk-management-framework
Recent Updates
- May 14, 2026: Colorado enacted SB 26-189, which repeals and replaces the earlier Colorado AI Act.https://leg.colorado.gov/bills/sb26-189https://www.mofo.com/resources/insights/260515-colorado-hits-reset-on-ai-regulation
- January 1, 2027: The revised ADMT framework is the operative effective date stated in the bill materials and practitioner commentary.https://leg.colorado.gov/bill_files/116432/downloadhttps://www.mofo.com/resources/insights/260515-colorado-hits-reset-on-ai-regulation
- Public legislative record updated in August 2026: The official Colorado bill page shows ongoing updates and posted bill materials, including versions last updated in July and August 2026.https://leg.colorado.gov/bills/sb26-189https://leg.colorado.gov/bill_files/116432/download
- No confirmed later amendment identified in the materials reviewed: Based on the retrieved sources, I did not see a later official amendment after the enacted 2026 bill package.https://leg.colorado.gov/bills/sb26-189https://leg.colorado.gov/bill_files/116432/download
- Important practical point: Some legal commentary notes the framework faced litigation risk shortly after enactment, so organizations should monitor the official Colorado legislature page and state guidance for implementation changes.https://www.buchalter.com/insights/colorado-rewrites-its-ai-law-what-employers-must-know-about-sb-26-189/https://leg.colorado.gov/bills/sb26-189
Related Regulations
- Colorado Privacy Act (CPA): The ADMT statute overlaps with privacy obligations where personal data collection, processing, and consumer rights are implicated; organizations should coordinate notices, vendor controls, and data inventories.https://cdle.colorado.gov/colorado-privacy-acthttps://leg.colorado.gov/bill_files/116432/download
- EEOC/Title VII employment discrimination rules: Employment uses of ADMT can create discrimination risk if screening or ranking has disparate impact or uses protected-class proxies.https://www.eeoc.gov/statutes/title-vii-civil-rights-act-1964https://www.eeoc.gov/youth/ai
- Fair Housing Act: Housing-related ADMT must also comply with federal fair housing restrictions on discriminatory effects in tenant screening and related decisions.[https://www.hud.gov/program_offices/fair_housing_equal_opp/fair_housing_act_overview](https://www.hud.gov/program_offices
Recent developments
- — Official Colorado General Assembly page for SB 26-189 detailing the **Colorado Automated Decision-Making Technology Act**, including its scope over “covered ADMT,” consumer notice obligations, developer documentation duties, and the **effective date of January 1, 2027**.[1] (source)
- — Revised bill text clarifying that, starting **2027-01-01**, developers of covered ADMT must provide technical documentation (intended uses, training data categories, limitations, human review instructions) to deployers and retain compliance records for at least three years.[2][7] (source)
- — Employment law analysis explaining how SB 26-189 **repeals and reenacts** Colorado’s prior AI Act, substantially **altering obligations for employers** that use AI/ADMT in employment decisions, including new notice, review, and recordkeeping requirements.[3] (source)
- — Practice guide overview highlighting that the new Colorado law **delays enforcement to 2027-01-01** and shifts regulatory focus from “High‑Risk AI Systems” to **Covered ADMT**, summarizing emerging compliance expectations for AI and ADMT deployers in Colorado.[14] (source)
- — Legal insight describing how Colorado has **rewritten its AI law**, replacing the 2024 AI Act with SB 26-189, shifting focus from “high‑risk AI systems” to **covered ADMT** and noting that enforcement is already subject to legal challenge, leaving the framework effectively **on hold**.[4] (source)
- — Client alert reporting that Governor Polis **signed SB 189** (SB 26-189) on 2026-05-14, significantly amending Colorado’s AI regime by narrowing coverage to ADMT that materially influences **consequential decisions** (e.g., education, employment, finance, insurance, healthcare, government services) and delaying effectiveness to **2027-01-01**.[5] (source)
- — Analysis emphasizing that Colorado’s amended AI law **substantially reduces employer obligations** compared to the prior AI Act and delays the effective date to **2027-01-01**, requiring clear notices to individuals, adverse action procedures with meaningful human review, and three‑year record retention for covered ADMT.[9] (source)
- — Compliance-focused industry page outlining SB 26-189’s **core operational impacts**: mandatory pre-use consumer notices when ADMT is used in consequential decisions, 30-day adverse outcome explanations and meaningful human review, plus detailed developer documentation obligations, emphasizing the law’s role as a **key AI/ADMT compliance driver** effective 2027-01-01.[11] (source)
- — Law firm commentary explaining that Colorado’s legislature **repeals and replaces** its earlier AI Act with SB 26-189, introduces comprehensive governance for ADMT in **consequential consumer decisions**, and pushes key compliance and enforcement obligations back to **2027**, giving industry additional implementation time.[8] (source)
- — Policy-focused blog describing legislative debates around SB 26-189, including **criticisms from stakeholders** seeking stronger transparency, shorter cure periods (currently extending to 2030), and a private right of action, illustrating mixed industry and advocacy reactions to the revised framework.[15] (source)
Related regulations
- Colorado Artificial Intelligence Act (SB 24-205) — Colorado, Superseded, effective 2026-06-30
- Brazilian Artificial Intelligence Act — Brazil, Proposed
- NIST AI Risk Management Framework (AI RMF 1.0) — United States, Active, effective 2023-01-26
- Artificial Intelligence and Data Act — Canada, Superseded
- Texas Responsible Artificial Intelligence Governance Act (TRAIGA, HB 149) — Texas, Active, effective 2026-01-01
- EU AI Act - Annex III High-Risk System Requirements (2 Dec 2027) — European Union, Upcoming, effective 2027-12-02
- EU AI Act - GPAI Model Obligations (2 Aug 2025) and Enforcement (2 Aug 2026) — European Union, Active, effective 2025-08-02
- EU Digital Omnibus on AI (Regulation (EU) 2026/1744) — European Union, Active, effective 2026-07-27
Put it into practice
- Generate the policy: Colorado AI Act policy generator (generatepolicy.com)
- Buy the policy pack: AI Risk Assessment Template (cyberpolicy.shop)
- Build it yourself: AI Risk Register Pack (ciso.diy)
Browse all regulations · Compliance deadlines · Latest updates