Colorado passed the nation’s first comprehensive artificial intelligence protection law, aimed at curbing discrimination against the public that could result from the technology’s use while still allowing AI entrepreneurship to flourish.
The law (Senate Bill 205), signed by Gov. Jared Polis (D) on Friday, includes mandates for AI creators and users of “high-risk” AI systems, which are systems that contribute to a consequential decision, like employment.
The bill doesn’t take effect until Feb. 1, 2026. Polis said that before then he wants to see it amended to ensure innovation won’t be stifled by the law.
Under the law, developers of high-risk AI systems must take reasonable steps to avoid inherent discrimination. They must disclose enough information about their systems to users so that users can conduct risk assessments. Creators must describe to the public the AI systems they have built or modified and how any potential inherent discrimination is controlled. If creators learn about the potential for discrimination after a system is in use, they have 90 days to inform the Colorado attorney general.
Businesses that use high-risk AI systems need to tell the public they are using such a system and how the risks are managed. They must implement a risk management policy and program, including a risk assessment of the system. Companies must also alert members of the public if AI is involved in making a major decision, such as hiring. The business must tell the attorney general within 90 days if it discovers discriminatory practices on behalf of the AI.
Small businesses with 50 or fewer full-time employees are exempted from maintaining a risk assessment program and informing the public about their use of AI.
The attorney general will enforce the law.
The law gives the public the right to know if an AI system was involved in making a major decision about them, to correct an incorrect outcome, and to appeal for a human to review the decision. Individuals must be able to opt out of any profiling for AI systems.
Laws against discrimination generally prohibit intentional discrimination, but the Colorado AI bill “deviates from that practice by regulating the results of AI system use, regardless of intent,” Polis wrote in a letter to lawmakers Friday. He urged the legislature to change the bill so the threshold for enforcement is “intentional discrimination.”
Polis also wants lawmakers to work with stakeholders to make sure “Colorado remains home to innovative technologies and our consumers are able to fully access important AI-based products,” he said.
No comments yet