The morning began with Denver’s mayor interviewing a monk about ethical artificial intelligence and ended several hours later with Gov. Jared Polis on a topic of great consternation for many in the technology industry: Colorado’s AI law.
The topic was bound to come up Tuesday at the second annual DenAI Summit, an AI conference hosted by the city of Denver. The state’s law, which Polis signed in May 2024, aimed to protect consumers from possible harms of AI. But pushback from the tech community over worries of stifling innovation led Polis to change his mind followed by two attempts by the state legislature to change the law.
Other than delaying its start to June 30, the law has not changed.
“The law’s central requirements for developers to use reasonable care and prevent algorithmic discrimination are widely criticized as vague. How can companies operating in Colorado avoid what seems to be an impossible and ambiguous legal standard and what are you doing to prevent this uncertainty from forcing high-growth AI startups to abandon our state?” asked Susie Loyacona, from the Daniels Fund.
Polis said he favors a national law, which came close under President Donald Trump’s tax cut and spending bill until the Senate removed the 10-year moratorium on state AI regulations. But any federal law is three-to-four years out.
“We need a pro-innovation, pro-growth AI policy,” Polis responded. “Absent that, again I think what states need to do and the way I would view any policy in Colorado is how we position Colorado as the best state to innovate in AI. … And that there’s a reason to come here from a legal perspective because you have the ability to reach consumers with lower costs and with greater predictability. That would be what I would seek in any changes to our AI law.”
Consumer advocates didn’t seem to be in the room for DenAI, an event geared toward civic leaders and tools that use AI to speed up services like permitting, ordering and operations. During testimony when Senate Bill 205 was underway, they spoke of the harm already caused to job seekers, renters or loan applicants who may unknowingly get rejected by AI systems that have been trained on biased or inaccurate data.
But that harm was touched on during the morning session when Denver Mayor Mike Johnston interviewed Tenzin Priyadarshi, a Buddhist monk and president and CEO of The Dalai Lama Center for Ethics and Transformative Values at the Massachusetts Institute of Technology.
“The notion of biased data has been overly spoken about in terms of transparency. But the cities and citizens need to get more involved in what technologies should city governments adopt,” Priyadarshi said. “Because that’s not always the first concern or the primary concern of either the designers or the companies that are selling them. It falls in the purview of city governments and state governments and federal governments.”
In fact, Priyadarshi said he showed up in Denver this week because of what the city was doing. “Denver is one of the few places that is actually openly talking about governance in AI,” he said. “Most times, cities and governments simply become consumers of these technologies. Very few of them are actually helping frame the policy around it.”
Coincidentally, California Gov. Gavin Newsom on Monday signed Senate Bill 53, the state’s safety law on artificial intelligence to protect consumers from potential harms of AI. It was considered a “diluted version” of an earlier bill vetoed last year that “mandated safety testing of AI and required companies to create a kill switch,” according to a New York Times report.
Cris Turner, who handles government affairs for Google and was on stage with Polis, said Google prefers the California law that passed calling it “a better bill to look at than the bill that’s currently in Colorado. But again, it’s still early days.”
The problem with any new law are the unintended consequences, he said. When laws are too vague, that impacts the budding industries and businesses that use AI. And for many small companies and local governments, AI is a tool that can be used for good — and bad.
“As you heard the governor say today, many of those harms can be addressed through existing regulation,” Turner said. “Because if this is a tool, bad folk are going to use it to be better at being bad, but we already have laws on the books. So how does AI apply to those laws versus creating some broad air quote AI law where it may not be necessary.”
Other panelists spoke about the challenge of setting rules in place when the rules are constantly changing. Eric Hysen, who joined SalesForce as its chief AI and transformation officer over the summer, had previously served as CIO at the U.S. Department of Homeland Security during the Biden administration and worked on the AI Safety and Security Board with tech and civic leaders.
“We came up with a framework that I think was a very good policy product but it already looks outdated when you look at how it defined the respective roles of cloud companies, AI model developers, AI deployers and others,” Hysen said. “The roles we’re playing are evolving more rapidly than I think we’ve seen in any other period of technological change.”
“Flexibility is key,” he said.