Read the text of Colorado’s new bill, which removes some of the stricter controls in the original bill’s framework.

The filing

Law360, “Colo. AG Agrees To Pause Enforcement Of Landmark AI Law,” reported May 4, 2026. Also Colorado SB26-189.

What happened

The Colorado Attorney General’s office will temporarily halt enforcement of the state’s AI law while lawmakers revise it, following litigation by Elon Musk’s xAI challenging the law on constitutional grounds. AG Philip Weiser agreed not to investigate or enforce against xAI for 14 days after the court rules on xAI’s motion for a preliminary injunction. Meanwhile, lawmakers are advancing SB26-189, which drops the strict “high-risk” compliance assessment for a transparency approach, removes mandatory bias audits, adds a 90-day correction period, and pushes implementation to January 2027. 
 
The original Colorado law aimed to ensure that consumers are clearly informed when AI systems influence important decisions affecting their lives. However, xAI’s complaint presents a different view, claiming that the law forces companies to move away from objective truth-seeking and instead promotes state-approved ideological views, especially on racial justice. The company argues that although Colorado officials describe the law as fighting algorithmic discrimination, it actually acts as a speech regulation that compels AI developers to adhere to controversial political orthodoxies. Additionally, the U.S. Department of Justice has expressed its concern, arguing that the law violates equal protection rights under the Fourteenth Amendment. 

Meanwhile, Colorado lawmakers are moving forward with SB26-189, a bill that marks a significant shift from the original plan. The new legislation drops the strict “high-risk” compliance assessment in favor of a transparency-oriented approach that prioritizes disclosure and human oversight options. For example, if an AI system rejects someone’s apartment application, the applicant must be informed about the AI’s involvement and provided a chance for human review. The bill also removes mandatory bias audits, which were considered especially burdensome, introduces shared liability for AI developers and deployers, includes a 90-day correction period for violations, and postpones the implementation date to January 2027. 

This enforcement pause and legislative recalibration have significant implications for AI governance across the country. Colorado’s experience highlights the difficulty states encounter in balancing consumer protection with innovation. The constitutional questions involved are likely to impact how other jurisdictions develop their AI regulations in the future.

Why it matters

Colorado was the bellwether for state AI regulation, so its retreat from strict compliance toward disclosure signals how other states may recalibrate under constitutional and federal pressure. 

JetStream’s take

The first comprehensive state AI law blinking is a reminder that the compliance ground is still moving, and that “wait and see” is itself a risk position. 

Source

2026a_189_signed-act by chelsea.stark.ctr