When Colorado replaced its groundbreaking AI law in May 2026, the move looked like a retreat from ambitious legislation. The state abandoned a unique framework that required companies to actively prevent algorithmic discrimination. This is the risk that automated systems will produce biased results when making decisions about people’s jobs, loans or health care based on race, gender, age or other protected characteristics.
Instead, lawmakers passed something far narrower on May 14, 2026, including a set of transparency requirements that dictate what companies must disclose to consumers.
I examine how AI and technology are transforming policymaking and democratic accountability. I also track state AI legislation using the US State AI Policy Tracker at the University of Denver Daniels College of Business.
Colorado’s back-and-forth on AI legislation shows a fundamental shift in the way state leaders think about governing AI. It also raises an unresolved constitutional question that could determine whether a state AI law survives a federal challenge.
Table of Contents
From prevention to disclosure
The original Colorado law, called the Colorado Artificial Intelligence Act, required companies to prevent harm. The replacement requires companies to inform consumers. This distinction may sound technical, but it reflects a significant shift in the prevailing philosophy.
The Colorado Artificial Intelligence Act is based on direct commitments. Developers and operators of high-risk AI systems—such as tools for screening job applicants, evaluating loan applications, or determining insurance eligibility—have had to establish risk management programs. The law also requires companies to conduct impact assessments and take proactive measures to avoid discriminatory outcomes. Essentially, the state told corporations: govern yourself and prove that you do.
The replacement bill, called the Automated Decision-Making Technology Act, takes a different approach.
Companies must notify consumers when automated technology plays a role in a subsequent decision. This may include a hiring decision or a loan denial. If this decision is against a consumer, the company has 30 days to explain how the technology contributed to the outcome.
Consumers can also request corrections of inaccurate personal data and meaningful human verification. The due diligence requirements, risk management regulations and impact assessments from the first version of the law no longer apply.
For Coloradans, the practical difference is significant. Under the old law, a company using AI to reject a loan application, for example, would need internal processes designed to detect discriminatory patterns before they reach the consumer.
Under the new law, that company must tell the consumer that AI was involved and explain its role afterwards. However, the company is not obliged to audit the system that led to the decision. The burden shifts from the institution using the technology to the person affected by it.
Equal protection out, free speech in
A federal lawsuit filed in April 2026 quickly moved the timeline for this law overhaul forward. Elon Musk’s company xAI has sued to block enforcement of the Colorado Artificial Intelligence Act, the original law.
Days later, the DOJ intervened in support of xAI. It was the first time that the federal government requested that a state AI law be declared invalid.
According to a report from 9News, the Justice Department backed a lawsuit led by Elon Musk’s xAI to block Colorado’s artificial intelligence regulations before they took effect in June 2026.
In its filing, the DOJ raised two constitutional arguments against the original Colorado law. The new law has neutralized the first argument, but may have exacerbated the second.
The first was an equal protection argument. The DOJ argued that the law’s anti-discrimination framework effectively forces developers to make race- and gender-conscious decisions about the behavior of their models. For example, companies would have had to test whether their AI systems produced different results for different population groups and adjust them if necessary.
This type of government-imposed classification is a matter of heightened scrutiny in the courts, meaning the government bears a heavy burden to justify it.
The new law completely eliminated this argument. The algorithmic discrimination provisions are gone, and there is nothing in the statute that requires developers to monitor their results or adjust for discriminatory patterns.
The second argument advanced by the DOJ was a First Amendment argument, specifically what is called “compelled speech.” The idea is simple: The First Amendment protects not only a person’s right to speak, but also the right not to speak. When a government requires a private company to deliver specific messages to consumers, that mandate must meet a constitutional standard. Courts ask whether the government has a good reason to force someone to say something they otherwise wouldn’t say.
xAI characterized the Colorado law’s disclosure and reporting requirements as precisely this type of forced communication.
In my opinion, this argument has been strengthened, not weakened, by the recast. The new Automated Decision-Making Technology Act provided little other than notice and disclosure requirements, so the essence of the new law is the government telling companies what they have to say to consumers.
By limiting the law in this way, the state legislature may have made the compelled speech argument more easily isolated and litigated, even if it eliminated the equal protection argument.
The executive order behind the lawsuit
The DOJ’s decision to intervene in a state-level litigation was no accident. It followed directly from Executive Order 14365, which President Donald Trump signed in December 2025.
According to a report from NBC News, President Donald Trump has signed an executive order that lays the groundwork for federal testing of the world’s most powerful AI systems before they are publicly released.
The order declared that a “minimally burdensome national policy framework” should govern AI and directed the DOJ to challenge state laws that contradict that vision. The order also directed the Commerce Department to publish an assessment of existing state AI laws and identify those it deems “burdensome.” It included a specific mandate to flag laws that compel disclosures that could violate the First Amendment.
This assessment was due to be completed by March 2026 but has not yet been published. If that list comes to fruition — and if Colorado’s new law ends up on it — it will signal that the administration sees this state not as a single problem to solve, but as a test case for a broader federal campaign against state AI regulation.
Dozens of states have introduced AI-related legislation in 2026, touching on the same areas that Colorado sought to regulate, including employment, lending, housing and health care.
What Coloradans should expect
Not much is changing in Colorado right now. A federal judge suspended enforcement of the original law in April 2026, and that suspension also applies to the replacement law. This means that none of its provisions can yet be enforced.
The stay will remain in effect for at least 14 days after the court’s decision on xAI’s request for an interim injunction, which is a court order that would prevent the law from coming into force while the case is decided. This request will not be made until 28 days after the completion of the State’s rulemaking process. So the timeline depends on the willingness of the attorney general, the courts and the DOJ to challenge a law that has already made significant concessions.
https://theconversation.com/why-colorado-replaced-its-ai-discrimination-law-with-a-transparency-requirement-that-the-feds-might-challenge-anyway-286517
