Legislative lowdown: Colorado enacts revamped AI law

In 2024, Colorado made history by becoming the first US state to pass legislation regulating AI use in the workplace, among other settings.

But history didn’t pan out as planned. The 2024 law was ultimately axed following pushback from members of the business community and governments, including the Trump administration and Elon Musk’s xAI.

Colorado Gov. Jared Polis signed a new bill on May 14 that repeals and replaces the 2024 legislation. Here’s what employers should know about that law, which will take effect on Jan. 1, 2027.

A more targeted approach to AI regulation. Colorado’s new AI law applies to certain types of “automated decision-making technology” (ADMT) involved in “consequential decisions,” including those related to an individual’s employment.

The new legislation is less onerous than its predecessor, as it removes a number of requirements. Employers, for example, will no longer have to conduct an “impact assessment” to determine whether the technology poses a risk of “algorithmic discrimination.” They won’t have to report discriminatory outcomes to the Colorado attorney general, either, nor review AI tools on an annual basis.

Rather than requiring employers to evaluate AI tools before they put them in place, this new law shifts the focus to what happens after they’re already being used, attorneys with law firm Jackson Lewis explained in a blog post. “Employers will be judged less on whether they performed the right assessments and more on whether they can articulate a clear, consistent, and non-discriminatory reason for each decision,” they wrote.

When it takes effect, employers in the state will have to do the following when using ADMTs:

• Disclose how the technology is being used to help make a “consequential decision”
• Disclose the role the technology played in cases where a “consequential decision” had an adverse outcome, within 30 days of the decision being made
• Allow employees to review data used in a “consequential decision” with an adverse outcome, and request to correct any inaccuracies in the data

Compromising on AI. The main sponsor of the 2024 bill, Colorado’s Senate Majority Leader Robert Rodriguez, described the new law as a “discrimination decision bill” and said he would have preferred a more comprehensive version, Colorado Public News reported.

Kjersten Forseth, legislative director for the Colorado AFL-CIO, said “there’s still work to be done to ensure Big Tech companies are transparent and accountable,” per the Colorado Sun, but added, “we’re encouraged that there’s now a path to hold developers accountable when their technologies result in discrimination and other harm.”

The Colorado legislature’s compromise comes as President Donald Trump is pushing back on state-level efforts to regulate AI. He specifically called out the 2024 Colorado law in a Dec. 11 executive order, positing that it could “force AI models to produce false results in order to avoid a ‘differential treatment or impact’ on protected groups.”

Still, the Trump administration recently indicated it’s open to regulating AI at the federal level. On June 2, the White House issued an executive order asking AI companies to give the federal government access to their models for up to 30 days before they release them to the public. The directive is not mandatory, however.