Earlier this year, Colorado state legislators passed a landmark artificial intelligence bill, SB 205 (Consumer Protections for Interactions with Artificial Intelligence). And although the term “landmark” is often overused, it’s certainly appropriate in this case. As many observers have indicated, the Colorado AI Act is the country’s first comprehensive AI bias law.
What does “comprehensive” mean, in this case? It means that this law doesn’t shy away from the the big issues. As attorneys at Venable LLP point out, “This landmark legislation, among the first of its kind in the United States, imposes rigorous standards on businesses employing ‘high-risk’ AI tools in critical sectors like employment, housing, finance, education, and healthcare.”
At the crux of the legislation is that concept of “high-risk” AI tools. Foley & Lardner LLP suggests that the Act “defines ‘high-risk AI systems’ as those that make or are a substantial factor in making ‘consequential’ decisions. It defines a “substantial factor” as a factor that assists in making a consequential decision, is capable of altering the outcome of a substantial decision, or is generated by an AI system.”
What does all that mean? In a nutshell, it means that when an AI tool is used in making a major decision, such as one affecting someone’s career, finances, health, or life, the business that has developed that tool, as well as the business using (or deploying) it, need to comply with certain provisions in the Act to ensure that the tool does not rely on biased data to make its decisions.
In the weeks before the Act was passed, there were many concerns on all sides about its perceived flaws. Many AI entrepreneurs thought it was too restrictive of their ability to effectively operate in Colorado without potentially punitive requirements. These requirements, while perhaps seen by many large tech companies as just additional legal hoops to jump through, were considered to be something of an existential threat to Colorado startups.
At the same time, other parties, such as Consumer Reports, thought the bill didn’t go far enough in providing protections for all the Coloradans who would be affected by AI in some of the most important interactions and transactions of their lives, such as applying for a job, purchasing a home.
However, over the course of several weeks, the bill went through some fairly significant modifications, and ultimately, it wound up on the Governor’s desk for his signature. On May 17, 2024, Gov. Jared Polis signed the bill, albeit with “reservations,” and thus the bill became law, the Colorado AI Act.
And yet, one of the most unusual things about the Act is that it will not go into effect until February 1, 2026, nearly 20 months after it was signed.
This 20-month delay allowed Polis to declare that despite his reservations, he thought that the Act was a good starting point and there would be sufficient time ahead for the Colorado General Assembly to amend it in ways that would further improve it.
In the months ahead, Colorado AI News will be taking a deeper dive into the Act, what it’s likely to mean for Coloradan citizens and businesses, and how it might be amended before it takes effect.
