Colorado retreats from sweeping AI bias law in shift toward lighter-touch rules

Colorado lawmakers are moving to roll back a landmark artificial intelligence law before it ever takes effect, underscoring how difficult it has been for U.S. regulators to sustain broad oversight of rapidly evolving technology.

A bill introduced in the state legislature would repeal and replace the 2024 measure signed by Governor Jared Polis, scaling back what had been the most comprehensive state-level attempt to curb algorithmic discrimination. The proposal, Senate Bill 26-189, would instead adopt a narrower approach focused on transparency and consumer recourse, according to the Colorado General Assembly.

The shift is drawing attention beyond Colorado, as officials in Europe and elsewhere weigh how far to go in regulating artificial intelligence.

From ambitious safeguards to pared-back rules

When Colorado passed Senate Bill 24-205 in 2024, it set out to tackle risks from “high-risk artificial intelligence systems” head-on. Companies would have been required to test systems for bias, reduce discrimination risks, and explain how AI shaped decisions affecting people’s jobs, housing, healthcare, and insurance.

But the law quickly ran into resistance. Industry groups argued the rules were too burdensome, and implementation was delayed amid mounting legal and political pressure ahead of its planned 2026 launch.

The new bill would strip out most of those proactive requirements. Instead, companies would only need to notify individuals when automated decision-making tools are used and provide more detailed explanations if someone receives an unfavorable outcome and asks for it.

Consumers would also be able to seek human review of such decisions, though the bill allows companies flexibility based on what is “commercially reasonable.”

Lawmakers call it a practical reset

State Senate Majority Leader Robert Rodriguez, who backed both the original and revised laws, said the new proposal reflects a compromise between competing priorities.

“This bill strikes an appropriate balance of protecting consumers while not being onerous,” Rodriguez told Axios.

He also described the measure as “more of a notice bill,” signaling a shift toward informing consumers rather than requiring companies to actively prevent harms, according to The Colorado Sun.

Senate President James Coleman, a co-sponsor, has said the legislation is the product of ongoing efforts to refine the state’s approach, according to local media reports.

A broader debate over how to regulate AI

The rollback highlights a deeper divide in how governments should regulate artificial intelligence.

Some researchers and civil society groups have argued that requiring companies to assess and mitigate risks before deploying AI systems is essential to prevent systemic bias. Others, particularly in the business community, have warned that such rules could slow innovation and create significant compliance burdens.

Colorado’s new approach leans toward a more familiar U.S. model—prioritizing transparency and allowing problems to be addressed after they arise—rather than the stricter, risk-based system adopted in Europe.

A narrow window for action

Lawmakers now face a tight timeline. The bill is set to be heard in committee, with the legislative session ending May 13, leaving little time to move it through both chambers.

Whatever the outcome, Colorado’s experience is already being watched as a test case. After briefly setting the pace for AI regulation in the United States, the state is now illustrating how quickly momentum behind sweeping rules can shift when political, legal, and economic pressures converge.

There’s a middle ground between leaving money in the bank and rolling the dice in crypto. Start with this free video on decentralized finance.