Utah Approves AI for Prescription Refill Process as States Test AI Governance Models
Utah has taken a significant step by approving a new Regulatory Mitigation Agreement (RMA) aimed at piloting AI in the prescription refill process. This move is part of a broader trend among states responding to the rapid adoption of AI technology, which has led to the introduction of numerous AI-related laws across various sectors.
As industry standards evolve, there is an emerging patchwork of regulations, driven by concerns that the speed of AI innovation may outpace necessary consumer protections for developers and users alike. Both Utah and Texas have established structured pilot frameworks that may address the AI Litigation Task Force‘s call for “minimally burdensome” oversight, thereby offering potential models for AI developers interested in testing high-risk applications within a regulated environment.
Challenges in AI Integration
A recent report indicates that business leaders are facing unprecedented challenges in integrating AI into their organizations effectively. In response, states like Utah are actively embracing pilot programs that allow collaboration with government agencies and stakeholders to manage risks associated with AI deployment.
Utah’s Regulatory Framework
Utah’s AI Policy Act (UAIPA), enacted in May 2024, marked the state as the first to regulate AI usage by organizations in consumer interactions. This law aims to enhance consumer protections while promoting responsible AI innovation. Key components of the UAIPA include:
- Transparency: Mandating consumer disclosure, especially concerning high-risk interactions involving healthcare data.
- Liability Clarification: Defining liabilities related to AI business operations.
- Innovation Enablement: Facilitating innovation through a regulatory sandbox, RMAs, and policy rulemaking by an Office of Artificial Intelligence Policy (OAIP).
Since the enactment of the UAIPA, Utah has entered into RMAs with various organizations, including:
- ElizaChat: An app designed to assist teens with mental health issues.
- Dentacor: A partnership focused on diagnosing specific dental conditions.
- Doctronic: A healthcare technology company aimed at streamlining the prescription refill process.
Understanding RMAs in AI
RMAs are structured agreements between participants, the OAIP, and state agencies designed to manage AI-related risks. While they do not completely shield from liability, RMAs allow AI developers and users to test their technologies in a controlled environment. The RMA with Doctronic exemplifies this approach, as it focuses on automating routine tasks related to prescription refills.
Key Features of the Doctronic RMA
The Doctronic RMA, which lasts for 12 months and spans 24 pages, includes:
- Schedule A: Plans to monitor and minimize risks associated with Doctronic’s technology.
- Schedule B: Details on use cases addressing clinician burnout and access to essential services, outlining how the AI system will process medication renewal requests.
- Schedule C: A list of 192 medications covered under the RMA alongside their associated conditions.
Evaluating AI Risk Mitigation Frameworks
The AI risk mitigation frameworks being utilized in Utah are among many proposed across various jurisdictions. A basic five-factor test for evaluating these frameworks includes:
- What is the AI tool? Understanding its type, models used, and legal implications.
- What is the use case? Assessing the sensitivity of the objective and the tolerance for error.
- What data is used? Evaluating the sources and potential risks associated with data.
- What outputs are generated? Identifying possible actions and their impacts.
- How accurate is the AI? Measuring accuracy and acceptable error levels.
Conclusion
For developers considering AI tools as the next big breakthrough, the RMA framework presents a structured approach to risk management. The UAIPA serves as a guide for those seeking clarity in deploying AI technologies within regulated environments, potentially shaping the future of AI governance.
