Colorado Rewrites Its AI Law Before It Takes Effect

Colorado’s legislature has passed a complete overhaul of its AI law, replacing Senate Bill 24-205 with Senate Bill 26-189, which now awaits Governor Jared Polis’ approval. The original law, enacted in 2024, aimed to regulate 'high-risk' AI systems used in employment and other high-stakes decisions, but faced criticism for its complexity and potential unintended consequences. The revised bill abandons the 'high-risk' framework, instead targeting automated decision-making tools that significantly impact consequential outcomes—such as hiring, promotions, or compensation. The new law broadens the scope to include any system processing personal data to generate predictions, recommendations, or classifications used in employment or consumer decisions. Employers must now assess whether their AI tools materially influence decisions about individuals, including job applicants and employees. The shift simplifies compliance by focusing on real-world impact rather than abstract risk categorization. Senator Robert Rodriguez, the original sponsor of SB 24-205, introduced the rewrite after delays and legislative debates stalled implementation. The bill passed the Colorado Senate with a 34-1 concurrence vote following House amendments, signaling bipartisan support for a more practical approach. If signed by Polis, the law will take effect January 1, 2027, repealing the original framework. Critics of the 2024 law had argued it created operational uncertainties, such as defining 'high-risk' systems or requiring technical visibility for third-party tools. The revised version eliminates these ambiguities by directly addressing how AI influences decisions, rather than preemptively classifying systems. This change aligns Colorado’s approach with emerging national trends in AI governance, prioritizing transparency and accountability in automated decision-making processes. The new law also clarifies that 'consumer' includes employees and job applicants, ensuring protections extend to all stages of employment. While the original bill mandated governance programs and impact assessments, the rewrite streamlines requirements to focus on measurable outcomes. This shift reflects broader concerns about AI fairness and the need for adaptable regulatory frameworks as technology evolves.