Demetri Giannikopoulos

Understanding Recent AI Regulations and Guidelines

Last Updated: January 29, 2024

The rapid adoption of artificial intelligence (AI) in healthcare has given rise to a flurry of legislative and guidelines measures, each aiming to shape the future of AI applications in clinical environments. Here we delve into the meaning behind the following four key initiatives and interpret their impact on healthcare AI moving forward:

1. Executive Order (EO) 14110: Guidance for AI Transparency and Safety

Overview: The EO, issued in October, is a broad order intended to provide a federalized approach to the safe and responsible use of AI. While not healthcare specific, it does task the Secretary of Health and Human Services (HHS), the Director of the National Institute of Standards and Technology (NIST), the Director of the Office of Science and Technology Policy (OSTP), and others with developing a position and response in alignment with federal priorities/guidance/values.  

Key Takeaway: The EO is as sensible as it is expansive. It emphasizes speed, transparency, safety and practical application of AI, not only in healthcare but also in social media, finance and other industries. It creates a broad framework to promote a pathway for transparency in AI development.

2. American Medical Association (AMA) Principles: Shaping Responsible AI Practices in Healthcare

Overview: The American Medical Association’s guidelines and principles are often seen as a standard bearer in healthcare. While they can’t enforce compliance, the principles provide crucial considerations for AI use while also emphasizing liability considerations of the use of AI applications in clinical practice.

Key Takeaway: These principles offer healthcare facilities and clinicians crucial considerations when using AI. They also set a precedent for accountability in AI application, highlighting the importance of responsible implementation and emphasizing enterprise-wide adoption for maximal healthcare utilization benefits. The AMA principles urge transparency from AI vendors in terms of regulatory approval status, clear description of any limitations or risks of use, and detailed information regarding data used to train the model. The AMA also makes recommendations about which party (physician, health system, or vendor) should be liable for clinical decisions related to AI utilization and AI-derived insights.

3. Evaluating Commercial AI Solutions in Radiology (ECLAIR): International Insights into AI Adoption in Radiology

Overview: Stemming from a 2021 European study, the ECLAIR guidelines are amongst the first codified recommendations for what to ask AI vendors, with emphasis on transparency, bias concerns and intended use of AI software implications in radiology settings.

Key Takeaway: ECLAIR acknowledges radiology’s position as the early adopters of healthcare tech, and offers guidance in evaluating AI vendors including in-depth risk assessment, clarification of algorithm design specifications, and so on. Its emphasis on transparency and data sharing agreements aligns with the evolving landscape of AI governance.

4. Health Data, Technology, and Interoperability (HTI-1) Final Rule: Defining AI in Healthcare Legislation

Overview: The HTI-1 Final Rule is closely aligned with EO 14110 and establishes rules for AI use in healthcare. It emphasizes the definition of AI, reclassifies actionable clinical decision support as decision support initiation (DSI) and provides a framework on how to assess and manage your AI based predictive DSI for certified health IT vendors and modules.

Key Takeaway: A huge takeaway of HTI-1 is that it clearly defines what AI is in the scope of legislation. One critical shift in language from the draft to the final rule clarifies the scope of AI, requiring it to be “supplied by” rather than  “enabled or interfaced with” certified healthcare IT vendors. This not only defines AI in legislation, but actively addresses implementation requirements for certified Health IT Vendors/Modules, and hints at its integration into broader healthcare standards.

Key Healthcare Timelines

EO 1440

  • February 2024 – Companies developing or demonstrating an intent to develop potential dual-use foundation models to provide the Federal Government, on an ongoing basis, with information, reports or records. Secretary of “HHS must establish an HHS AI Task Force that shall, within 365 days of its creation, develop a strategic plan that includes policies and frameworks[…]including research and discovery, drug and device safety, healthcare delivery and financing, and public health.”
  • August 2024 –  Establish guidelines and best practices, with the aim of promoting consensus industry standards, for developing and deploying safe, secure and trustworthy AI systems


  • December 2024 – Clinical Decision Support Certification must be updated to Decision Support Intervention certification for certified Health IT developers 
  • December 2024 – FHIR Endpoints must be published by certified health IT vendors
  • December 2025 – Certified Health IT vendors must conform to the USCDI v3 scope of data for interoperability standard and certification

Explore the Latest AI Insights, Trends and Research

Demetri Giannikopoulos