EDITORIAL: A flawed AI bill Polis might have to sign
Well, here’s another fine mess ruling Democrats at the legislature have gotten us into — this time, thanks to their bungled attempts at regulating artificial intelligence. Their latest such effort is now on its way to the governor’s desk.
It likely will do more harm than good for Colorado. Indeed, the state would have been better off if lawmakers never had meddled in the first place.
Yet, Gov. Jared Polis now might have no choice but to sign it.
It seems the legislature couldn’t resist being “first in the nation” to put “guardrails” on the development and deployment of artificial intelligence in 2024. Lawmakers insisted they were heading off “algorithmic discrimination” by AI as it’s used throughout the consumer economy, but no one seemed sure what constitutes such discrimination — or whether it exists.
The legislation was ready to backfire before it left the Capitol building, with its prime sponsor, Senate Majority Leader Robert Rodriguez, D-Denver, acknowledging as much. Polis, who signed the complicated and misguided measure into law — albeit with a delayed implementation date — began publicly denouncing the policy before his ink had dried.
He and others rightly feared the new law’s regulatory overkill would put the chill on Colorado’s tech sector by scaring off investment in AI, tech’s new driving force. They must have realized the law set unattainable standards for development and use of AI-driven software that would miss the mark and succeed only in squelching the technology, chasing off AI innovators to less oppressive states. An economic analysis released last year by Colorado’s Common Sense Institute said the law “could impose severe economic costs on the state if it goes into effect …”
Or, even before it went into effect. In February, AI pioneer Palantir Technologies, with a market capitalization estimated at $328 billion, announced it would move its headquarters from Denver to Miami. At least one factor prompting the move reportedly was the pending AI law.
The governor had nudged the legislature over a year ago to postpone the law’s implementation again — to next month — and convened a working group of lawmakers, consumer advocates, tech industry leaders and others to work out compromise legislation to replace the law.
What emerged only a week ago was Senate Bill 189, which was approved by both chambers and sent to the Governor’s Office this week, barely in time for today’s legislative adjournment.
The finished product that will face the governor is still, at best, problematic. Although it’s a marked improvement over the 2024 law, the new measure still touches pretty much every industry in the state — and doubtless will serve as a tripwire for a torrent of litigation.
It requires, as Colorado Politics reported, developers of “automated decision-making technology” to disclose the technology’s intended use to deployers and users. Among other provisions, the bill includes user disclosure requirements so that if a consumer alleges that the automated decision-making was used to make a consequential decision resulting in an “adverse outcome,” deployers — lenders, retailers, landlords; the list goes on — will be required to provide a description of the role the technology played in influencing that decision.
In other words, it’s wrapped in red tape. It’s also an open invitation to lawsuits by an endless procession of disgruntled consumers who don’t like a decision — even if it’s the same decision that would have been made by an ordinary human.
All of which means it still has considerable potential to scare off AI developers and users to more welcoming climes.
But Polis has painted himself into a corner. If he doesn’t sign the bill, the far worse 2024 measure will take effect.
If safeguards are warranted for rapidly proliferating AI, a cohesive congressional solution makes more sense than a state-by-state, regulatory patchwork. Polis has said repeatedly he prefers that tack. So does the president, who in December issued an executive order blocking states from mandating their regulations for artificial intelligence.
It would have been so much easier to do nothing — and watch AI flourish in our state.
Instead, our politicians had to “do something” — and open a can of worms.




