If your organization uses AI to screen resumes, rank candidates, or flag employees for promotion, Colorado’s regulatory landscape is coming into focus. On October 6, 2026, the Colorado Department of Law released an interim draft of rules (4 CCR 904-6) implementing the state’s automated decision-making technology (ADMT) law and its new chatbot law. With key rules due by January 1, 2027, the draft is the clearest preview yet of how Colorado will regulate AI in workforce decisions.

What are Colorado’s new ADMT and Chatbot Safety laws?

Colorado’s ADMT law (Senate Bill 26-189) regulates automated decision-making technology: tools that use personal data to score, rank, sort, recommend, or predict something about an individual to help make an important (“consequential”) decision about that person, such as resume screeners, candidate-ranking tools, and assessment platforms. The Chatbot Safety Act (House Bill 26-1263) regulates public-facing chatbots, but internal workplace chatbots and business productivity or training tools are excluded (the draft rules confirm that workforce-only tools behind a login are outside the law).

What is the status of the proposed rules?

The Attorney General proposed rules on August 11, 2026, to implement both laws and fill in details the statutes leave open. The October 6 interim draft responds to public feedback. A second round of comments was requested, and a public hearing was held on October 26, 2026.

Does the ADMT law apply to employers?

Yes. Hiring and other employment decisions are expressly covered, and protected “consumers” include employees and Colorado-resident job applicants. Coverage may also reach anyone whose opportunity in Colorado is evaluated, so multistate employers should not assume out-of-state applicants for Colorado jobs are excluded.

What changed in the October 6 interim draft?

The changes that matter most to employers are:

  • A presumption that AI output counts. If a decision-maker sees a tool’s score, ranking, recommendation, or similar result and the decision matches it, the tool is presumed to have “materially influenced” the decision. Employers may rebut that presumption, but doing so can be burdensome.
  • Hiring and promotion examples. An internal promotion tool that scores candidates is presumed to influence the decision even if the manager also interviews everyone and reviews their files. The same applies to a tool that ranks applicants from “strongest” to “weakest” without a numeric score.
  • Using a staffing firm doesn’t shift the duty. If a staffing company’s AI tool produces a shortlist you use to hire, you are responsible for compliance, though you can contract with the firm for assistance.
  • More detailed rejection notices. Notices must specifically explain the tool’s role, give the principal reasons for the outcome, disclose any score or inference used and the underlying data, identify any automatic-disqualification factor and, if information was incomplete, say what is missing.

What must employers do to comply?

  • Tell people up front. Before using a covered tool, give applicants and employees a clear, conspicuous, plain-language, and accessible notice that the tool is being used and how to obtain more information.
  • Explain negative outcomes. If the tool contributes to a rejection or other negative decision, explain in writing within 30 days, through every channel you normally use to communicate with the employee or applicant. The notice must include information on, for example, what was decided, how the tool was involved, how to obtain more information, and the individual’s rights.
  • Offer a fix and a human second look. If asked, explain how to correct inaccurate data and provide meaningful human review where commercially reasonable. The draft would also require pausing the adverse outcome, where possible, while data is corrected. Do not retaliate against anyone exercising these rights.
  • Use qualified human reviewers. Reviewers, who may be internal or external under the draft rules, must be trained, have authority to override the decision, and consider the underlying evidence rather than defer to the tool.
  • Keep records. Keep records showing compliance for at least three years. Records may include covered ADMT version identifiers, changelogs, and documentation of material mitigation changes.

What are the liability and enforcement risks?

  • Discrimination claims. The law creates no new private right of action, but AI-influenced decisions can still be challenged under existing discrimination law, including the Colorado Anti-Discrimination Act. Both the employer and the vendor can be liable, and vendor contracts cannot shift responsibility for your own discriminatory conduct.
  • Who enforces. Only the Attorney General can enforce the law. Until January 1, 2030, the Attorney General generally must give notice and 60 days to cure before suing, where a cure is possible, except for knowing or repeated violations.

What should employers do now?

With the rules taking effect January 1, 2027, employers should:

  • Whether developing AI and algorithmic tools or deploying such tools developed by a third party, understand their features, limitations, potential use cases, as well as how permitted users might obtain, evaluate, and make decisions concerning output. Inventory the tools used in hiring, promotion, pay, and discipline, and identify which ones materially influence decisions.
  • Get documentation from vendors (due starting January 1, 2027), revisit staffing and recruiting contracts, and build notice templates and workflows that meet the applicable deadlines.
  • Ensure developer service agreements address several key areas including, without limitation, transparency, risk allocation, service level, change management, privacy, cybersecurity, and compliance obligations.
  • Train human reviewers and set up documentation of independent judgment, such as interview notes, to help rebut the presumption.
  • Establish a process for tracking changes to tools the organization has deployed.
  • Update record retention guidelines to ensure the organization is keeping compliance records for at least three years.
  • If applicable, evaluate obligations in other states (e.g., CA, CT, IL, NY) which may create obligations distinct from those under Colorado’s law.

If you have questions about compliance with the Colorado statute or similar AI regulation, contact a member of our Artificial Intelligence and Innovation Group or the Jackson Lewis attorney with whom you regularly work.

Photo of Joseph J. Lazzarotti Joseph J. Lazzarotti

Joseph J. Lazzarotti is a principal in the Berkeley Heights, New Jersey, office of Jackson Lewis P.C. He founded and currently co-leads the firm’s Privacy, Data and Cybersecurity practice group, edits the firm’s Privacy Blog, and is a Certified Information Privacy Professional (CIPP)…

Joseph J. Lazzarotti is a principal in the Berkeley Heights, New Jersey, office of Jackson Lewis P.C. He founded and currently co-leads the firm’s Privacy, Data and Cybersecurity practice group, edits the firm’s Privacy Blog, and is a Certified Information Privacy Professional (CIPP) with the International Association of Privacy Professionals. Trained as an employee benefits lawyer, focused on compliance, Joe also is a member of the firm’s Employee Benefits practice group.

In short, his practice focuses on the matrix of laws governing the privacy, security, and management of data, as well as the impact and regulation of social media. He also counsels companies on compliance, fiduciary, taxation, and administrative matters with respect to employee benefit plans.

Privacy and cybersecurity experience – Joe counsels multinational, national and regional companies in all industries on the broad array of laws, regulations, best practices, and preventive safeguards. The following are examples of areas of focus in his practice:

  • Advising health care providers, business associates, and group health plan sponsors concerning HIPAA/HITECH compliance, including risk assessments, policies and procedures, incident response plan development, vendor assessment and management programs, and training.
  • Coached hundreds of companies through the investigation, remediation, notification, and overall response to data breaches of all kinds – PHI, PII, payment card, etc.
  • Helping organizations address questions about the application, implementation, and overall compliance with European Union’s General Data Protection Regulation (GDPR) and, in particular, its implications in the U.S., together with preparing for the California Consumer Privacy Act.
  • Working with organizations to develop and implement video, audio, and data-driven monitoring and surveillance programs. For instance, in the transportation and related industries, Joe has worked with numerous clients on fleet management programs involving the use of telematics, dash-cams, event data recorders (EDR), and related technologies. He also has advised many clients in the use of biometrics including with regard to consent, data security, and retention issues under BIPA and other laws.
  • Assisting clients with growing state data security mandates to safeguard personal information, including steering clients through detailed risk assessments and converting those assessments into practical “best practice” risk management solutions, including written information security programs (WISPs). Related work includes compliance advice concerning FTC Act, Regulation S-P, GLBA, and New York Reg. 500.
  • Advising clients about best practices for electronic communications, including in social media, as well as when communicating under a “bring your own device” (BYOD) or “company owned personally enabled device” (COPE) environment.
  • Conducting various levels of privacy and data security training for executives and employees
  • Supports organizations through mergers, acquisitions, and reorganizations with regard to the handling of employee and customer data, and the safeguarding of that data during the transaction.
  • Representing organizations in matters involving inquiries into privacy and data security compliance before federal and state agencies including the HHS Office of Civil Rights, Federal Trade Commission, and various state Attorneys General.

Benefits counseling experience – Joe’s work in the benefits counseling area covers many areas of employee benefits law. Below are some examples of that work:

  • As part of the Firm’s Health Care Reform Team, he advises employers and plan sponsors regarding the establishment, administration and operation of fully insured and self-funded health and welfare plans to comply with ERISA, IRC, ACA/PPACA, HIPAA, COBRA, ADA, GINA, and other related laws.
  • Guiding clients through the selection of plan service providers, along with negotiating service agreements with vendors to address plan compliance and operations, while leveraging data security experience to ensure plan data is safeguarded.
  • Counsels plan sponsors on day-to-day compliance and administrative issues affecting plans.
  • Assists in the design and drafting of benefit plan documents, including severance and fringe benefit plans.
  • Advises plan sponsors concerning employee benefit plan operation, administration and correcting errors in operation.

Joe speaks and writes regularly on current employee benefits and data privacy and cybersecurity topics and his work has been published in leading business and legal journals and media outlets, such as The Washington Post, Inside Counsel, Bloomberg, The National Law Journal, Financial Times, Business Insurance, HR Magazine and NPR, as well as the ABA Journal, The American Lawyer, Law360, Bender’s Labor and Employment Bulletin, the Australian Privacy Law Bulletin and the Privacy, and Data Security Law Journal.

Joe served as a judicial law clerk for the Honorable Laura Denvir Stith on the Missouri Court of Appeals.

Read more about Joseph J. Lazzarotti
Show more Show less