In a significant development for the artificial intelligence industry, xAI has filed a federal lawsuit to block the enforcement of Colorado’s new artificial intelligence law before it takes effect. The legal action, targeted at Colorado Attorney General Philip Weiser, seeks to halt the Consumer Protections for Artificial Intelligence (CPAI) law, which imposes stringent requirements on developers of “high-risk” AI systems to protect consumers from algorithmic discrimination. The complaint can be seen in detail here.
The lawsuit highlights six constitutional claims, focusing on First Amendment and Equal Protection concerns. xAI contends that developing an AI model constitutes an “expressive act,” and that the CPAI law compels the company to redesign its systems. Under the precedent set by 303 Creative v. Elenis and Moody v. NetChoice, the First Amendment argument claims that Colorado is mandating changes to expressive content, invoking a strict scrutiny analysis. Under this rigorous judicial review, the law must serve a compelling state interest through the least restrictive means. xAI asserts that CPAI fails this criterion.
Furthermore, the lawsuit argues that the law’s key terms are vague, including “historical discrimination.” It also criticizes the law’s Equal Protection exemption for AI systems used to boost diversity or remediate historical discrimination, alleging a race-based double standard without adequate justification.
Additional challenges include the law’s extraterritorial reach. xAI argues that since CPAI applies whenever a Colorado resident is affected by an AI system, regardless of location, it unconstitutionally regulates out-of-state transactions. The Dormant Commerce Clause of the Constitution bars states from regulating commerce beyond their borders, and xAI posits that CPAI violates this by governing AI interactions happening outside Colorado based merely on a resident’s involvement.
Set to take effect on June 30, the CPAI law defines high-risk systems as those that, when deployed, significantly influence a decision. It requires developers to disclose publicly how systems are assessed and mitigated for bias, and mandates notification to the attorney general within 90 days of any discovered or potential “algorithmic discrimination.” Violations, viewed as unfair trade practices under Colorado’s Consumer Protection Act, carry a penalty of $20,000 per violation, with exclusive enforcement privileges granted to the attorney general.
Interestingly, AG Weiser has expressed concerns about the law’s efficacy in the past. As noted in Reuters coverage, his office has yet to comment on the lawsuit.
This lawsuit represents another chapter in the expanding field of AI-related legal battles. Previous cases have seen settlements related to AI’s societal impacts, including a lawsuit against Google and Character.AI concerning a teenager’s 2024 suicide, and a class-action settlement involving Anthropic over the use of pirated materials for model training.