In recent discussions surrounding AI governance, the events unfolding in Colorado reveal a significant shift in regulatory approach. The state, which once positioned itself as a pioneer in artificial intelligence legislation, is now reevaluating its stance on these regulations.
Colorado currently has a Democratic trifecta and a Democratic triplex, controlling the offices of governor, secretary of state, attorney general, and both chambers of the state legislature.
The state’s recent retreat provides context as news emerges that the Trump administration is backing off its plans to contest state-level AI regulations. The situation in Colorado illustrates how the influence of the AI industry can lead states to hesitate in their regulatory efforts.
By Stefani Langehennig, an Assistant Professor of the Practice in the Business Information & Analytics Department at the University of Denver’s Daniels College of Business. Previously, she served as a Lead Data Scientist at ICF, advising U.S., U.K., and E.U. agencies on predictive analytics and policy evaluation. Originally published at The Conversation.
In May 2024, the passage of the Colorado Artificial Intelligence Act garnered significant national attention. It was the first of its kind in the United States and aimed to regulate “high-risk” AI systems across various sectors to prevent potential real-world harm.
Governor Jared Polis signed the legislation with reluctance. Yet, less than a year later, he is advocating for a federal pause on state-level AI regulations. Colorado lawmakers have delayed its enactment until June 2026 and are exploring ways to repeal and replace portions of the original law.
Facing pressure from the technology sector and lobbyists, lawmakers are also considering practical challenges related to the cost of implementation.
What Colorado decides in the coming months will determine whether its initial actions serve as a model for other states or a cautionary tale about the difficulties involved in regulating emerging technologies.
As someone who studies the impact of AI and data science on policymaking and democratic accountability, I am especially interested in the lessons that Colorado’s early efforts might provide to state and federal legislators.
The First State to Act
In 2024, Colorado legislators opted not to await federal guidance on AI policy, as Congress continued to struggle with legislative gridlock stemming from political polarization. Instead, states began taking charge of AI governance.
The Colorado AI Act defined “high-risk” AI systems as those that impact significant decisions in employment, housing, healthcare, and daily life. Its ambitious goal was to provide consumer safeguards against algorithmic discrimination while fostering innovation.
This proactive stance from Colorado isn’t surprising given its reputation for embracing technological advances and its expanding AI sector. The state took a leading role in AI governance, drawing inspiration from global models like the EU AI Act and privacy laws such as the California Consumer Privacy Act of 2018. With an effective date set for Feb. 1, 2026, lawmakers have ample opportunity to refine their definitions, establish oversight structures, and enhance compliance capabilities.
Upon its passage in May 2024, the law was lauded as a major breakthrough by policy analysts and advocacy organizations. In fact, other states, such as Georgia and Illinois, attempted to draft legislation mirroring Colorado’s AI Act, although those efforts ultimately did not succeed. The Future of Privacy Forum described it as the “first comprehensive and risk-based approach” to AI accountability. This nonprofit organization focuses on research and advocacy concerning data privacy and emerging technologies.
Legal experts, including attorneys general nationwide, noted that Colorado set a strong legislative framework that other states could emulate in the absence of federal standards.
Politics Meets Process, Stalling Progress
Despite the praise, translating a bill into effective action poses its own challenges.
Once the legislation was signed, tech companies and trade groups cautioned that the act could impose considerable administrative burdens on startups, potentially stifling innovation. In his signing remarks, Polis emphasized the risk of “a complex compliance regime” that might hamper economic growth. He urged legislators to reconsider portions of the bill.
Polis convened a special legislative session to reevaluate parts of the law. Several bills were introduced aimed at amending or postponing its implementation. Meanwhile, industry leaders advocated for narrower definitions and extended timelines, while consumer advocacy groups pushed to uphold the law’s protections.
Other states are observing the developments closely, reconsidering their own AI legislation. For instance, Gov. Gavin Newsom stalled California’s ambitious AI bill, facing similar concerns, while Connecticut’s legislation faltered amid a veto threat from Gov. Ned Lamont.
Colorado’s initial advantage is becoming precarious. The same boldness that placed it at the forefront makes the law susceptible to changes—especially as governors can veto, delay, or amend AI legislation in response to shifting political landscapes.
From Big Swing to Small Ball
To maintain its leadership in AI governance, Colorado could benefit from an approach of “small ball,” or incremental policymaking, which emphasizes gradual improvements through ongoing monitoring and iterations.
This approach would not only account for ambitious goals but also delve into the practical implications of implementation. Key steps could involve clarifying what constitutes high-risk AI applications, defining compliance responsibilities, launching pilot programs for regulatory testing, and developing impact assessments to monitor the effects on innovation and equity. Additionally, engaging developers and community stakeholders in establishing norms and standards would be crucial.
This incremental approach is not a retreat from earlier objectives but a recognition of realism. Successful policies often emerge from gradual refinement rather than sweeping changes. For example, the EU’s AI Act is being implemented in stages, highlighting the importance of a phased approach, as noted by legal scholar Nita Farahany.
Effective governance of complex technologies necessitates a willingness to iterate and adjust. This principle has also been evident in data privacy, environmental regulations, and social media oversight.
In the early 2010s, social media platforms operated without constraints, yielding both benefits and new challenges. It was only after significant research and public pressure that governments began to regulate content and data practices.
Colorado’s AI legislation may mark the beginning of a similar path: an initial, imperfect step that encourages further learning, revision, and the establishment of standards across states.
The primary challenge lies in finding a workable balance. Regulations must protect individuals from unjust or unclear AI-driven decisions without imposing burdens that deter businesses from innovating or adopting new technologies. With its vibrant tech sector and practical policy approach, Colorado is poised to exemplify that balance through its commitment to incremental and accountable governance. By doing so, it has the potential to transform a disjointed start into a comprehensive framework for other states looking to manage AI responsibly.
