Colorado Lawmakers Advance AI Law Rewrite as Session Ends

Colorado lawmakers are racing to finalize a rewrite of the nation's first comprehensive artificial intelligence regulation before the legislative session closes. Senate Bill 189, introduced Friday in the Colorado Senate, represents the state's third attempt to revise the 2024 law that has faced intense pushback from both tech companies and consumer advocates.

The new measure strips out the original law's most burdensome requirements, including mandatory bias audits and detailed disclosures to the Colorado Attorney General's Office. Under Senate Bill 189, companies creating and using AI systems would no longer have to explain how their algorithms make decisions on hiring, loans, or housing. They would still need to notify consumers when AI is being used for consequential decisions and provide an appeal mechanism.

Senate Majority Leader Robert Rodriguez, a Denver Democrat who authored the original 2024 legislation, called the new bill "more of a notice bill" that maintains the basic premise of consumer notification. The measure delays the effective date from June 2026 to January 2027, giving businesses additional time to adapt (a problem that has plagued users for years, frankly).

The bill largely follows recommendations from a working group convened by Governor Jared Polis. Developers would need to provide deployers—governments, universities, hospitals, and banks—with information on intended use, harmful applications, training materials, and known limitations. Deployers must retain this information for at least three years and give consumers ways to request human review of AI-driven decisions.

Stakeholders remain split on the compromise. Dennis Dougherty, who leads the AFL-CIO in Colorado, said the coalition People for Responsible Technology is "cautiously optimistic" about the bill. Bryan Leach, CEO of Denver-based shopping app Ibotta, called it "a marked improvement over the original bill" but criticized the expiration of the right-to-cure provision after three years.

That provision would allow developers and deployers to resolve violations without facing civil penalties under the Colorado Consumer Protection Act. Rodriguez defended the sunset clause, noting it's "not an eternal get-out-of-jail-free card." The legislation also clarifies liability between developers and deployers, requiring penalties to be allocated based on relative fault.

The rewrite comes amid escalating legal challenges. Elon Musk's xAI company filed a lawsuit on April 9 alleging the existing law is unconstitutional. The U.S. Department of Justice joined the suit, arguing the law jeopardizes America's position as "the global AI leader." A federal court issued a temporary restraining order on April 27, blocking enforcement until June 30.

According to Colorado Public Radio, the original law created an uproar among tech companies over concerns it was too stringent. Consumer advocates felt it was too weak. Finding common ground has confounded the legislature for more than two years.

The court order temporarily blocks the state from taking enforcement action while protecting employers from investigation or penalties for alleged violations. The court said xAI and the DOJ should file new motions within 28 days after lawmakers pass amended legislation or if the state AG issues final regulations implementing the law.

With the legislature wrapping on May 13, time is growing short. According to Fisher Phillips, bill co-sponsor Rep. Brianna Titone believes the legislature will pass the rewrite before time runs out. The original 2024 law, SB24-205, was approved by Governor Polis on May 17, 2024, and originally set to take effect February 1, 2026.

For businesses operating in Colorado, the physical reality of compliance means auditing every AI-assisted tool involved in hiring, performance evaluation, or compensation decisions. Companies need to understand what each tool does, how it influences decisions, and what data it uses. If vendors can't answer these questions, that's important information in itself.

Whether the compromise satisfies anyone remains questionable. Rodriguez acknowledged that if everyone is unhappy, it likely means a good compromise has been reached. But the real test comes when companies actually implement the requirements and consumers try to navigate the appeal processes.

The question isn't whether the law will pass. It's whether the streamlined framework provides meaningful protection or just creates a compliance checkbox that companies can tick without changing how they build or deploy AI systems. Whether users actually pay for it remains the real question.

Artūras Malašauskas is an AI Systems Integrator with 20+ years of production-grade web engineering experience. He has designed, shipped, and scaled enterprise Python/PHP systems for logistics, SaaS, and public-sector clients. For the past year, he has focused exclusively on AI integrations: deploying open-source LLMs, building generative media pipelines (image, audio, video), and engineering multi-agent workflows for real production environments. His standard: reproducibility, security, cost-efficient inference—no vaporware. He documents and evaluates emerging AI tooling, separating verified capabilities from marketing noise. Technical editor at: muza-ai.eu, ai-verslas.lt, ai-naujinos.lt Connect on LinkedIn