In an effort to revise one of the most scrutinized artificial intelligence regulations, Colorado legislators are proposing new rules governing AI’s role in decisions impacting employment, housing, and service access.
The proposed legislation aims to replace the current statute, SB24-205, enacted in 2024, with revised guidelines designed to balance industry concerns while continuing consumer protection. According to a summary of the bill, Senate Bill 24-205 initially established protections for consumers engaging with AI systems but now will be repealed and reinstated with additional stipulations concerning automated decision-making technology used in significant decisions.
The new bill, SB26-189, targets ‘consequential decisions’ within sectors like employment, education, housing, lending, insurance, healthcare, and government services. It pertains to tools that handle personal data for generating predictive or ranking outputs affecting these areas.
This legislative shift follows sustained opposition from the AI industry regarding Colorado’s 2024 law, which mandated companies to identify and mitigate bias in decisions related to hiring, lending, and housing. The original law was criticized for imposing excessive responsibilities on firms involved in developing and deploying AI within the state. In April, Elon Musk’s AI firm xAI filed a lawsuit against the state to prevent the enforcement of this regulation. This legal challenge received support from the U.S. Department of Justice, which sought to intervene in favor of xAI.
“I believe they present a legitimate case concerning the burdens Colorado’s policy would impose on these corporations,” Cody Barela, a partner at Armstrong Teasdale, previously commented to Decrypt. “The impact this has on these companies compared to the delay it might cause in AI development could be a stronger argument.”
Under the new legislation, developers must provide documentation detailing their systems’ operations, data usage, and known limitations, along with notifications for significant updates. Companies employing these systems are required to inform consumers about AI involvement in decisions, clearly explain adverse outcomes, permit individuals to access and amend their data, and allow requests for human review. Developers will also need to supply comprehensive system documentation, disclose the data they utilize, outline any limitations, and notify companies of material changes.
If approved, this bill would become effective on January 1, 2027. Meanwhile, legislators in other states such as New York and California are exploring similar regulations targeting AI systems, while the Trump Administration seeks to preempt state laws.