Colorado’s AI Act Rewrite Moves Forward: A New Draft for Employers
Colorado is moving forward with a major rewrite of its artificial intelligence law. A new draft bill, introduced by Senator Rodriguez on April 27, 2026, changes how employers must handle hiring decisions when using automated tools. Instead of focusing on whether a system is “high-risk,” the proposed law focuses on how decisions are made and how they affect people. This shift means employers need to look closely at their hiring workflows, not just their policies.
From Concept to Legislation
The earlier proposal for Colorado’s AI Act introduced a new way of thinking. Instead of labeling AI systems as “high-risk,” it asked a simpler question: Does the technology materially influence a big decision, like hiring or housing?
The April 27 draft keeps this approach. It applies when automated decision-making tools are used in areas like employment, credit, insurance, healthcare, or government services. This covers a wide range of tools—even those not traditionally seen as AI—that still shape outcomes.
What the New Draft Changes
The original Colorado AI Act required companies to set up risk management programs, run impact assessments, and build ongoing governance systems. The new draft removes those heavy requirements.
Instead, it focuses on what happens at the moment of decision-making. Employers must:
- Provide notice when automated tools are used in important decisions.
- Explain how the tool played a role in any negative outcome.
- Allow people to ask for corrections to inaccurate data.
- Offer meaningful human review of automated decisions.
This change reduces the need for formal governance programs but doesn’t remove compliance duties. For employers, the focus shifts from policies to workflows—how hiring decisions are structured, what role automated tools play, and how those decisions are explained to candidates.
A Clearer Liability Framework
The April 27 bill also introduces a more structured way to assign responsibility. It ties liability to how a system is designed, configured, and used. It also limits how much parties can shift liability to each other in contracts. This affects vendor agreements and indemnification clauses.
For employers using third-party tools, this is a big deal. It changes how you structure vendor relationships and understand responsibility when automated tools influence a decision. The bill also makes clear that using AI doesn’t replace existing legal duties, like anti-discrimination and consumer protection laws.
Where Questions Remain
Even with a more developed draft, some key questions are still open. The biggest one is: What does “materially influences” mean? The bill uses this standard to trigger its rules, but how it applies in real-world hiring workflows is unclear. For example, if a background check, an interview, and an AI score all play a role, when does the tool cross the line from helping to deciding?
There are also practical questions about how this bill works with other laws, like the Fair Credit Reporting Act (FCRA). Employers already follow FCRA rules for background checks, including sending pre-adverse and adverse action notices. The new draft adds more disclosure requirements but doesn’t fully explain how to align them with existing FCRA processes. This could mean layering new notices on top of old ones, creating complexity for both employers and candidates.
Finally, the distinction between developers and deployers is clearer but still tricky. In hiring workflows that rely on third-party providers for data or scoring, deployers may depend on those providers to meet disclosure and documentation rules.
What This Means for Employers
Colorado’s direction is becoming clearer. The state is moving away from regulating AI as a standalone technology and toward focusing on how decisions about people are made. That shift matches how employers actually use these tools.
But it also changes where compliance lives. The question is no longer: “Is this tool AI?” The question is: “Is this tool shaping the outcome?”
For employers, that means:
- Take a close look at your hiring workflows.
- Identify where automated tools are used and how they influence decisions.
- Make sure human review is not just available but meaningful.
- Review vendor contracts to understand who is responsible for what.
Colorado’s AI Law Is Still Evolving
The April 27 draft is a big step toward a more targeted and practical framework. But important details still need to be worked out, especially around disclosure rules and the “materially influences” standard. The timeline for enforcement is also still flexible as policymakers refine the law.
For employers, the takeaway is simple: Regulation is not slowing down. It’s becoming more focused on how technology and decision-making intersect. Colorado moved early in this space, and now it’s helping define what comes next. Compliance will live less in policy documents and more in the structure of the hiring process itself.
Comment