Colorado Working Group Reaches Consensus on AI Act Revisions
On March 17, 2026, Colorado Governor Jared Polis announced a significant development regarding the controversial Colorado AI Act. A working group, consisting of industry experts and civil rights advocates, successfully reached a unanimous agreement on a framework designed to amend the existing legislation.
Background
The working group was formed in response to ongoing disagreements that emerged after a special legislative session in August 2025 failed to produce viable fixes for the AI Act. Operating behind closed doors since October 2025, the group focused on crafting a solution that would protect consumers while fostering innovation.
Framework Overview
The newly proposed framework aims to regulate AI technology without hindering its development. It primarily focuses on developers and deployers of AI, referring to entities responsible for creating and implementing AI systems in various business contexts. The regulations emphasize the use of AI in connection with consequential decisions, which include:
- Educational enrollment or opportunities
- Employment opportunities
- Residential real estate transactions
- Financial and lending decisions
- Insurance underwriting and claims
- Healthcare services
- Eligibility for public benefits
Notification Requirements
Under the proposed framework, AI developers must inform deployers about how the AI technology functions in relation to consequential decisions. They are also required to disclose known risks and instances where the AI should not be employed. Furthermore, if AI leads to an adverse decision, such as an application rejection, the deploying entity must provide:
- A description of the decision and the AI’s role
- Information on the personal data used in the decision process
- A process for individuals to request a human-led review of the decision
Regulatory Timeline and Enforcement
The framework mandates that the Colorado Attorney General’s (AG) Office create rules by December 31, 2026. These rules will cover disclosure requirements post-adverse decisions, the process for human-led reconsiderations, and consumer rights to correct inaccurate data. Enforcement will be the sole responsibility of the AG’s Office, which can impose civil penalties for violations. Notably, there will be no private right of action allowed.
Liability and Developer Protections
To alleviate concerns surrounding developer liability, the framework specifies that blame for violations will be allocated between developers and deployers based on their level of fault. Developers can also avoid liability if deployers use AI in unintended or unapproved ways.
Future Outlook
As the Colorado legislative session progresses, it is anticipated that the framework will be officially adopted, amending the Colorado AI Act. While specifics of the AG’s rules remain to be seen, this scaled-back framework is designed to address many criticisms of the prior legislation, which has faced scrutiny since its enactment in 2024.
