Legal Series: Everything You Need to Know About Colorado’s Proposed AI Law

This article is part of an ongoing legal series designed to provide insight and practical guidance on current and emerging workplace compliance issues. These insights shared by lawyers are based on their interpretation of existing regulations and proposed changes, and intended for informational purposes, not to be regarded as legal advice.  

 

Uses of artificial intelligence (AI) programming continue to grow, as AI platforms offer new ways to complete more tasks in less time. While the federal government has been encouraging almost unfettered development, many states, including Colorado, have passed regulations related to the use of AI technologies in the workplace. In 2026, Colorado passed a new AI law that requires employers to take affirmative steps whenever they use AI-based programming tools to make consequential decisions. Shortly after the law passed, Elon Musk’s company X.AI challenged the law in court, joined by the federal government.

The federal court paused enforcement of the Colorado AI law while the lawsuit proceeds, and the Colorado government is still deciding how to respond. Outsolve is here to help you stay on top of developments like the Colorado AI Act to ensure you stay in compliance with the law. Subscribe to our weekly newsletter to get updates delivered to your Inbox.

What Does the Colorado AI Act Do?

Colorado Governor Jared S. Polis signed an updated Colorado AI Act from the Colorado General Assembly into law on May 14, 2026. The law repeals and replaces a previous attempt at a comprehensive Colorado AI law, which the state passed in 2024 before a judge blocked its implementation. That law required employers to establish risk management programs, conduct regular impact assessments, and establish anti-discrimination safeguards.

The updated regulations focus on developers and deployers of AI technology. Deployers are those who use the software, and the law particularly applies to employers who use AI technology to materially influence a consequential decision. Employers who rely on such technology must provide detailed information to anyone affected by its use, offer a structured review process with human oversight for those affected, and retain details for three years about decisions that use AI technology.

What Is Automated Decision-Making Technology?

The Colorado AI Act refers to the AI-based programming it regulates as automated decision-making technology (ADMT). ADMTs include any technology that processes personal data and produces an output. ADMT outputs may include, for example:

• Predictions,
• Recommendations,
• Classifications,
• Rankings, or
• Scores.

If a technology relies on computation to generate an output that you use to help you make a decision, it may be ADMT. For example, you might use a tool to rank and compare how candidates performed on a pre-employment screening.

What Does It Mean for ADMT to Materially Influence a Consequential Decision?

The current Colorado AI regulation applies specifically to uses of ADMT that “materially” influence a “consequential” decision. For the law’s purposes, a consequential decision relates to:

• An educational opportunity,
• Employment opportunities,
• Leases or purchases of real estate,
• Financial or lending services,
• Insurance benefits,
• Healthcare services, and
• Essential government services and benefits.

When it comes to employment opportunities, consequential decisions are those that relate to eligibility, compensation, and other terms.

The law states that the use of ADMT materially influences a decision if it is more than a “de minimis” factor. In other words, the technology materially influences your decision if you genuinely factor its output into the decision-making process.

What Must Employers Do to Comply?

Employers who use ADMT must:

• Disclose their use of ADMT before using it to make a consequential decision,
• Inform employees and applicants about how they can obtain additional information related to the decision-making process,
• Provide a plain language description of the decision and the role the ADMT played in it to anyone affected by an adverse outcome within 30 days of the decision, and
• Retain records explaining decisions for three years.

When you use an ADMT to make a decision that qualifies as an adverse outcome, your obligations increase. An adverse outcome denies something, revokes it, or changes a person’s compensation, access to work, or employment eligibility. For example, the decision to demote, fire, or not hire someone may be an adverse outcome.

Individuals who experience adverse outcomes have the right to request specific things from the employer following that decision. Employers who use ADMT to influence an adverse outcome must be prepared to:

• Provide instructions for requesting personal data and correcting inaccuracies in the data the ADMT uses,
• Allow the individual to correct that information, and
• Have a human being meaningfully review the decision.

Employers who use ADMT may be liable for any discriminatory decisions the ADMT makes.

Federal Intervention: X.AI v. Weiser

Before Governor Polis signed the new Colorado AI Act into law, Elon Musk’s X.AI company challenged it in federal court. The AI company claims that the Colorado law violates the Equal Protection Clause of the 14th Amendment.

The federal government joined the lawsuit, meaning U.S. government attorneys are now also challenging the Colorado law. Its intervention in the case aligns with its AI Litigation Task Force, a group the administration created through a December 2025 executive order to disrupt state-level efforts to regulate AI.

While the case is still in early stages, the court ordered the government to hold off on implementing the Colorado law. The next stages of the case may depend on whether the Colorado General Assembly decides to amend the law or allow the Colorado government to defend its law in court.

What Principles Can HR Professionals Follow?

The Colorado AI law is centered around efforts to prevent AI technologies from discriminating in violation of state and federal law, like discrimination based on race, sex, or sexual orientation. While the specifics of the law might shift, it signals a growing desire to ensure that AI tools do not perpetuate discrimination.

Understanding how biases enter generative calculations is critical to use such technologies effectively. Because these technologies train on human data, they can reflect and even amplify biases.

 For instance, the ADMT identifies names that don't match the expected demographic for a position. It may conclude that the person is not an ideal candidate for the job simply because they failed to meet the pattern—an implicit bias. Researchers have documented such biases among employers in the past. A human with an understanding of biases should regularly review ADMT decisions to identify and ameliorate these risks.

Moving Forward with AI

The legal landscape for AI-based technologies is constantly evolving. The Colorado AI Law signals a shift in how states and the federal government may approach the issue moving forward. Staying on top of changes to laws and regulations across the country may require flexibility as we figure out how to ethically use technologies without perpetuating explicit or implicit biases that may have influenced the training data.

Outsolve is here to help HR professionals and business owners as we navigate the ever-shifting field.

OutSolve

Founded in 1998, OutSolve has evolved into a premier compliance-driven HR advisory firm, leveraging deep expertise to simplify complex regulatory landscapes for businesses of all sizes. With a comprehensive suite of solutions encompassing HR compliance, workforce analytics, and risk mitigation consulting, OutSolve empowers organizations to navigate the intricate world of employment regulations with confidence.