With the reluctant stroke of Governor Jared Polis’ pen, Colorado this month became the primary US state to pass a law that explicitly regulates using artificial intelligence – a milestone that supporters say is an imperfect start line for establishing oversight of an emerging industry.
The state's latest law generally addresses the danger of discrimination in corporations' use of AI while requiring a basic level of transparency.
“I think Colorado's AI law represents a major shift in the way we approach AI oversight in the United States,” said Duane Pozza, a Washington, D.C.-based attorney specializing in emerging technologies and artificial intelligence. “…This is the first law that tries to regulate AI – or at least certain types of AI – more broadly and really put a lot of requirements on what are defined as high-risk AI systems.”
Starting in early 2026, the state would require certain corporations that use AI to make “consistent” decisions to reveal the technology's use and purpose to consumers, job applicants and others who interact with the technology. The law is designed to assist Coloradans who could also be screened by an AI tool after applying for a job, financial service, educational program or home or apartment.
Under the law, an applicant being screened by an AI tool will probably be informed that they’re coping with a machine and will probably be told the explanation why. If someone is rejected for a job or housing, they can even be given an evidence.
Developers of such AI tools can even must disclose more information in regards to the systems, resembling how they’re tested for bias. The measure broadly goals to limit the flexibility of AI systems to discriminate against certain people or groups, supporters said, even though it doesn’t change existing discrimination law.
“Depending on how you program the decision-making, (the AI) could exclude certain names, it could exclude certain people based on their race. In some of these things, bias is inevitable,” said Democratic Sen. Robert Rodriguez of Denver, who sponsored the bill.
“All we're asking,” he added, “is that companies notify anyone who interacts with it and that they do risk assessments and other things just to show that they're updating (the AI) when they see damage. Because that's what's going to happen.”
Colorado's law and similar attempts to pass laws in other states have run into conflict on many fronts, including between civil rights groups and the tech industry. Some lawmakers are wary of getting involved in a technology few understand, and governors are afraid of being seen as outsiders and scaring away AI startups. Polis expressed concern in regards to the stifling of AI innovation.
The states have discussed many other bills this 12 months that relate to narrower areas of AI, resembling using Deepfakes in elections or Making pornography.
When it involves more comprehensive regulations, measures much like Colorado's have failed in states like Washington and Connecticut, whose proposal served as a model for Colorado's first version. Another bill in California has survived thus far.
The latest law in Colorado is probably going one in all the primary comprehensive AI laws on the planet, said Rep. Brianna Titone, a Democrat from Arvada who introduced the bill together with Rodriguez and Democratic Rep. Manny Rutinel. The European Union will has also approved AI regulations.
Senate Bill 205 was repeatedly rewritten on its way through the state Capitol. Due to opposition from the tech industry, rumors circulated that it will be vetoed — until the governor's office announced in a press release Friday evening, May 17, that Polis had signed it.
Still, he expressed doubts: In a press release accompanying his signature, the governor wrote that he was “concerned about the impact this bill could have on the technology industry and AI developers.” He said he hoped lawmakers would improve the measure before it fully takes effect on February 1, 2026.
Rodriguez, the Senate majority leader, said he believes no further changes to the bill are needed next 12 months, but he and Titone each said additional AI regulations will undoubtedly be needed within the years to return.
This 12 months, the legislature passed an accompanying invoice This would expand an existing task force investigating artificial intelligence.
“Extremely modest” requirements?
When the bill was introduced, it was met with scepticism across the political spectrum. Numerous corporations, industry associations and worker groups made official statements in the course of the debate in Parliament; none of them spoke in favour of the bill.
Trade unions feared that the planned law could possibly be used to bypass existing discrimination protection measures. Consumer advocates called for stricter regulations. Technology and industry associations rejected the law because it will decelerate a growing industry.
These groups argued that the federal government should take the lead to make sure a unified approach.
Following the changes, employees' and consumer associations largely abandoned their open opposition and expressed greater satisfaction with the bill, although not overly enthusiastic.
“It contains some basic disclosure provisions that would at least shed some light on the shadowy world of AI-driven decisions,” said Matt Scherer, senior policy adviser for employees' rights and technology on the Center for Democracy and Technology in Washington, DC.
“In addition, the bill's requirements are extremely modest,” he said. “And I would say even the disclosure provisions fall far short of what public interest groups have been calling for. But at least it's a good foundation.”
Scherer said the bill provides exemptions for smaller businesses and for corporations that need to protect information they consider trade secrets. He also argued that the bill's enforcement provisions – which might ultimately allow Colorado's attorney general to take motion against an organization that violates the law – ought to be strengthened.
Arguments against a national approach
Meanwhile, technology corporations and industry associations proceed to oppose the bill. Some opponents, including the national Consumer Technology Association, sent letters to Polis urging him to dam the bill.
Doug Johnson, the association's vice chairman for emerging technology policy, expressed concern in an interview Wednesday that the bill would restrict a brand new technology and industry.
“We're in a brand new era of primordial soup,” said Logan Cerkovnik, founding father of Colorado-based Thumper.ai, of the AI field. “Overly restrictive legislation that forces definitions on us and limits our use of technology as it emerges will simply hurt innovation.”
Johnson and others argued that Congress should take the lead in a federal approach.
“Overall, policy efforts here need to focus on achieving a balanced and proportionate AI policy that puts rules and guardrails in place at the national level,” Johnson said. “… We don't need a fragmented, state-by-state approach.”
Rodriguez and Titone each said they didn’t need to wait for motion from a slow-moving Congress, especially since artificial intelligence appeared to have moved from science fiction to reality over the past 12 months.
Scherer rejected the “patchwork argument” that technology corporations use to avoid stricter regulations.
And besides, he argued, Colorado's law is not going to put an iron grip on AI.
“My conclusion is that it's a foundation to build on,” he said. “And it's definitely not a high-water mark to lower. It's a floor, not a ceiling.”
image credit : www.mercurynews.com