Revamped Colorado AI law targets ‘consequential’ HR decisions, takes effect in 2027

Revamped Colorado AI law targets ‘consequential’ HR decisions, takes effect in 2027

Dive Brief:

  • Colorado employers using artificial intelligence and similar automated technologies for certain “consequential” employment decisions — such as hiring and compensation determinations —  must provide notices that they are using those tools under a revised version of the state’s AI law signed by Gov. Jared Polis on Thursday.
  • Senate Bill 26-189 repeals the state’s previous law, which was set to take effect in June. Its protections apply to Colorado residents, those who have access to or eligibility for an opportunity in Colorado, or who are evaluated in a consequential decision by someone doing business in Colorado.
  • Employers using covered tools that “materially influence” decisions which lead to an adverse outcome for individuals must provide a description of the tool’s role in the decision, alongside other required disclosures, within 30 days of the decision. The law is effective Jan. 1, 2027 and applies to all decisions made on or after that date.

Dive Insight:

Colorado’s second attempt at AI legislation follows several months of debate and legal action challenging its first attempt, known as Senate Bill 24-205. State officials up to and including Polis expressed a desire to revisit the original law after its passage, citing concerns about the potential effects it would have had on innovation and competition.

Several parties opposed SB 24-205, including Elon Musk’s xAI, which sued Colorado’s attorney general last month to block enforcement of the law. xAI took particular issue with the first law’s mandate that organizations deploying AI tools take reasonable steps to avoid “algorithmic discrimination,” a term the company alleged to be unconstitutionally vague.

SB 26-189 scraps all references to algorithmic discrimination and also sets out a new definition of the technologies it seeks to regulate.

Whereas SB 24-205 applied to “high risk artificial intelligence systems,” the new law defines a broad category of “automated decision-making technology,” or ADMT, which includes any tool that processes personal data and uses computation to generate output, including “predictions, recommendations, classifications, rankings, scores, or other information used to make guide, or assist a decision, judgment, or determination” about an individual.

SB 26-189 also outlines several technologies that are not included under its ADMT definition, and tools are generally excluded if they are not used to make a consequential decision. Such decisions do not include “low-stakes or routine decisions” which do not materially influence eligibility for, selection for, dental of, compensation for or access to employment and similarly covered opportunities.

The law specifies that it does not cover instances in which an ADMT is used to summarize, organize or present information for human review and does not produce a score, ranking, recommendation, classification, prediction or other inference that materially influences an outcome or decision.

Individuals who experience adverse outcomes resulting from consequential decisions made by an ADMT may request instructions for requesting personal data and correcting factually inaccurate or materially inaccurate personal data used in the decision. They also may formally request an opportunity for “meaningful human review and reconsideration” of the decision.

SB 26-189 specifies that Colorado’s attorney general will issue regulations to clarify and implement its requirements on or before the 2027 effective date. If the attorney general finds a violation has occurred, the organization deploying the ADMT will have a period of 60 days to cure the violation, if possible, before the attorney general may bring legal action. The attorney general may forgo the cure period requirement if the organization is found to have knowingly or repeatedly violated the law.

In a press release, Polis said the new law represented a collaborative effort between consumers and various other stakeholders that would preserve Colorado’s role in the growing AI industry.

“This is a big step in the right direction for Colorado, and a model for the rest of the country,” Polis said in the press release. “Replacing the old law that hasn’t taken effect yet will boost Colorado innovation and entrepreneurship.”

The news comes just days after the state’s legislature passed a separate bill prohibiting employers from discriminating against workers by setting individualized wages using AI or similar tools. 

In response to SB 26-189’s passage, affected employers should map their AI tools and engage with vendors, Fisher Phillips attorneys said in an analysis. The attorneys also advised employers to note the law’s three-year recordkeeping requirement for relevant compliance documents and to watch for pending legal challenges to the revised law.