Introduction

In May 2026 Colorado replaced its pioneering AI law, the Colorado Artificial Intelligence Act, with a narrower version called the Automated Decision‑Making Technology Act. The original statute required developers of high‑risk AI systems to create risk‑management programs, conduct impact assessments, and actively prevent discriminatory outcomes in hiring, lending, and other consequential decisions. The revision discards these obligations and instead focuses on telling consumers when AI was used and providing a post‑decision explanation within thirty days.

Legal Shift and Its Implications

The new law’s transparency duties mean a company must merely notify a consumer that an automated tool contributed to a decision and, if the consumer challenges the outcome, explain the tool’s role. It no longer obligates firms to audit their models for bias before the decision is made. This shift transfers the burden of proof from the institution to the individual affected, reducing the state’s ability to enforce anti‑discrimination safeguards. Consumers still retain rights to request data corrections and meaningful human review, but the proactive safeguards that once required companies to monitor and adjust for unequal impact are gone.

Potential Federal Challenge

Shortly after the original law was passed, Elon Musk’s company xAI sued to block its enforcement, and the Department of Justice intervened. The DOJ’s initial argument centered on equal‑protection concerns, claiming the law forced developers to engage in race‑ and sex‑based classifications, which would trigger heightened judicial scrutiny. By removing the anti‑discrimination provisions, the new law eliminates that equal‑protection claim but sharpens the First Amendment challenge. The DOJ now contends that compelling companies to deliver specific disclosures amounts to forced speech, a claim that is likely to attract rigorous constitutional analysis.

What Colorado Residents Can Expect

Because a federal judge issued a nationwide stay on the original law, the revised statute is also unenforced while the stay remains. The stay will stay in effect until at least fourteen days after a court rules on xAI’s request for a preliminary injunction, which the state can file only after completing its rule‑making process. This legal uncertainty means that, for the time being, Colorado businesses continue operating under the old requirements, but any future enforcement of the transparency‑only regime could leave consumers less protected against biased AI outcomes.

Conclusion

Colorado’s decision to replace a robust anti‑discrimination framework with a limited transparency rule marks a retreat from proactive AI governance. While the law still aims to inform consumers, it removes essential safeguards that could prevent systemic bias. The strengthened First Amendment challenge suggests the revised act may face a formidable federal legal battle, leaving the state’s AI policy in a state of flux and its residents awaiting clearer protections.