Two New AI Laws, Two Different Directions (For Now)
Recently, two distinct approaches to AI legislation have emerged in the United States, particularly in Colorado and Utah. While Colorado’s legislature has rejected amendments to its AI Act, Utah has enacted amendments to enhance consumer protection while fostering innovation. This article analyzes both states’ legislative actions and their implications for AI use.
Key Takeaways
- Colorado legislature rejects amendments to the Colorado AI Act (CAIA).
- Proposed amendments sought multiple exemptions, exceptions, and clarifications.
- Utah legislature enacts amendments that include a safe harbor for mental health chatbots.
- Utah’s safe harbor provision includes a written policy and procedures framework.
Colorado’s Rejected Amendments
In Colorado, amendments to the AI Act were submitted to the legislature with high expectations. However, all proposed amendments were surprisingly rejected. This decision sets the stage for the CAIA to take effect on February 1, 2026, marking a significant milestone in regulating AI consumer usage.
The rejected amendments included technical changes, such as exempting certain technologies from the definition of “high risk” and creating exceptions for developers who disclose system model weights. Additionally, non-technical changes aimed to eliminate the duty of a developer or deployer of a high-risk AI system to protect consumers with reasonable care, which proved untenable.
Some amendments, like providing exemptions for systems below specific investment thresholds, were seen as reasonable. Their rejection indicates that extraordinary circumstances would be needed to delay the effective date of the CAIA.
Utah’s AI Amendments
In contrast, Utah has taken a proactive approach with its AI Policy Act (UAIP). The recent amendments include a regulatory sandbox for responsible AI development, aiming to balance consumer protection and innovation. These amendments are influenced by guidelines published for mental health therapists regarding AI use.
The guidance document released by Utah’s Office of AI Policy outlines potential benefits and risks associated with AI. It emphasizes the necessity of informed consent, disclosure, data privacy, and continuous monitoring in mental health therapy.
Among the most significant changes is the establishment of a safe harbor for mental health chatbots. This provision protects developers from litigation due to alleged harm caused by their chatbots, provided they adhere to specific requirements, including developing a written policy outlining the chatbot’s purpose and capabilities.
Final Thoughts
The divergent legislative outcomes in Colorado and Utah highlight the varying strategies states are adopting in response to the challenges posed by AI. While Colorado’s rejections signal a cautious approach, Utah’s amendments reflect an eagerness to embrace innovation while ensuring consumer safety.
As AI continues to evolve, maintaining a high level of AI literacy may become essential for both developers and users, ensuring that the technology can be harnessed responsibly.
