Just 53% of Colorado’s original AI Act survived its latest legislative gutting. That’s right. A law that was supposed to protect you from algorithmic discrimination in jobs, housing, and healthcare? It’s been neutered.
Look, when Colorado passed its AI Act back in 2024, it was a big deal. A genuine first. The kind of thing that makes you sit up and think, ‘Okay, maybe someone’s finally thinking about the actual people here.’ Gov. Jared Polis even signed it, albeit with some reservations. But that initial flicker of hope? It’s been systematically extinguished.
This week, the Colorado Legislature, bless their hearts, decided to pass SB 26-189. And what does this masterpiece of legislative backpedaling do? It repeals significant chunks of that original AI Act and, surprise, pushes the effective date back even further. Polis is apparently thrilled to sign this one too.
This isn’t just a minor tweak. This is a demolition. Gone are the duty of care for developers to, you know, not discriminate. Poof! The requirement for deployers to actually have risk management programs? Vaporized. Impact assessments? Reporting to the AG? All casualties of this legislative bloodbath.
So, what’s left? A faint echo. Companies still have to disclose when an AI makes a ‘consequential decision’ about you. And you get some limited rights—like correcting inaccurate data or, in some cases, a human might consider your appeal. But here’s the kicker: many of the companies most likely to use this stuff are exempt anyway. It’s like putting a speed limit on a highway but letting most of the cars ignore it.
And who’s enforcing this watered-down version? The Attorney General. And guess who can’t sue you directly if your rights are violated? You. Individuals have zero standing in court. So, if an AI screws you over, tough luck. You can complain, maybe, but don’t expect to see anyone in a courtroom unless the AG decides to pick up your case. This is almost as good as no law at all.
This whole saga is made even more amusing by the federal government’s involvement. The Trump administration, always eager to play whack-a-mole with state-level regulation, has been trying to squash this. Then there’s xAI, Elon Musk’s outfit, filing a lawsuit claiming the Act violates the First Amendment and Equal Protection. Apparently, telling companies they can’t discriminate is an infringement on free speech. Go figure. SB 26-189 conveniently removes many of the provisions xAI was complaining about, effectively making their lawsuit moot. Talk about strategic legislative maneuvering!
What’s truly fascinating is how quickly the conversation shifted from ‘consumer protection’ to ‘regulatory hurdles’ and ‘innovation stifling.’ This legislative session saw attempts to amend the law back in 2025, including a special session called by Polis himself. The original Act, a supposed bulwark against AI harms, became a prime target for anyone who felt it inconvenienced their profit margins or their ability to deploy AI without much oversight.
This isn’t just about Colorado. This is a canary in the coal mine for AI regulation across the US. If a state can pass a seemingly strong law and then have it so thoroughly dismantled, what hope do we have for consistent, meaningful protections elsewhere? The pressure from industry groups and the federal government (especially with administrations that are, shall we say, less concerned with tech regulation) is immense. It makes you wonder if the primary goal here was ever truly to protect citizens, or if it was just to check a box for ‘doing something about AI’ before rolling over.
Who’s Making Money Here?
This is the million-dollar question, isn’t it? The original AI Act, with its stringent testing and risk management requirements, might have cost some companies time and money. Developers and deployers would have had to invest in better auditing, more strong testing, and potentially redesigning some of their algorithms to ensure fairness. That’s money spent, not money made.
By gutting these provisions, Colorado has effectively lowered the barrier to entry for using AI in high-stakes decision-making. Companies can now proceed with less scrutiny, fewer upfront costs, and a lower risk of immediate legal challenges. This directly benefits the bottom line for any entity looking to automate processes and make decisions at scale without significant overhead for compliance and safety.
It also benefits the lobbyists and legal firms that have been working to shape this legislation. The constant back-and-forth, the special sessions, the legal challenges – these all create billable hours. And for the tech companies themselves, a weaker regulatory environment means faster deployment and less friction in bringing AI-powered products to market. Less regulation equals more opportunity to capture market share and generate revenue.
Why Does This Matter for Developers?
For developers working on AI systems, this means the pressure to build ethically and responsibly might be reduced, at least from a legal standpoint in Colorado. While many developers will continue to strive for best practices out of personal conviction or company culture, the legal mandate to avoid algorithmic discrimination or conduct thorough impact assessments is significantly diminished. This could lead to a more lax approach to fairness and bias testing if there’s no longer a clear legal imperative. It also means that the legal risks associated with deploying potentially biased systems are considerably lower in the state, shifting the burden of proof and potential penalties significantly.
What’s the Future of AI Regulation in the US?
This constant back-and-forth in Colorado is a microcosm of the larger struggle to define AI regulation in the United States. On one hand, you have states and advocacy groups pushing for consumer protections and ethical guardrails. On the other, you have industry lobbying for minimal interference and federal agencies sometimes leaning towards a more hands-off approach to foster ‘innovation.’ The xAI lawsuit against Colorado’s original Act illustrates the legal challenges inherent in state-level regulation, particularly when it intersects with First Amendment claims. It’s likely we’ll see more such legal battles and continued legislative flux at both the state and federal levels. The path forward is far from clear, and it will likely involve a complex interplay of court decisions, ongoing legislative efforts, and evolving industry standards, making it difficult to establish a consistent national framework anytime soon.
🧬 Related Insights
- Read more: ChatGPT Ate My Trade Secret: Two Court Rulings That Could Reshape AI Confidentiality
- Read more: Judges’ Secret Legislative History Revival: A Backdoor for Smarter Law Reading
Frequently Asked Questions
What does the revised Colorado AI Act require companies to do?
The revised law primarily mandates disclosures to individuals affected by AI-driven consequential decisions and grants limited rights for correction and human review in some cases. However, many high-impact users are exempt.
Will this law protect me from AI discrimination in Colorado?
While the law aims to provide some protections, many key requirements like duty of care and risk management programs were removed, and significant entities are exempt, weakening its overall protective capacity.
Can individuals sue companies for violating this AI law?
No, individuals do not have the right to sue companies directly for violations of their rights under this law; enforcement is exclusively handled by the Attorney General.
