The American Telemedicine Association on Wednesday released updated policy principles outlining how artificial intelligence should be deployed across healthcare settings, adding new guidance on validation, monitoring and data security as AI adoption accelerates.
The updated framework builds on principles first published in October 2023 and reflects work by the ATA's member-led AI Work Group, which positions the principles as a practical guide for providers, developers and policymakers navigating AI's role in telehealth, virtual care, digital health and hybrid care delivery.
The revisions emphasize self-regulatory practices, including real-world performance validation, continuous monitoring and ongoing improvement of AI systems used in clinical and operational environments.
A central theme of the framework is accountability. The ATA says responsibility for AI systems should be aligned with how those systems are designed, deployed and used, and should reflect their impact on clinical outcomes.
The principles call for clearly defined roles among technology developers, infrastructure providers and healthcare organizations to ensure that accountability is shared appropriately based on each party's level of control over system behavior.
Transparency and explainability are also emphasized. The ATA recommends patients be informed when they are interacting with AI or when AI materially influences their care.
Developers are encouraged to disclose the intended purpose of AI systems and the type of patient data analyzed, while avoiding repetitive or excessive disclosures that could lead to consent fatigue.
The guidance allows for different levels of transparency depending on whether AI is used in direct patient interactions or in background administrative functions.
Addressing bias is another key focus; the ATA states that AI systems used in healthcare should be routinely evaluated for bias and potential disparities in access or outcomes.
The principles support ongoing assessment and governance rather than static fairness requirements, acknowledging that standards may evolve over time. The organization also encourages the use of appropriate training data and policies that incentivize organizations to identify and correct AI-related issues when they arise.
On regulation, the ATA calls for harmonized, risk-based frameworks rather than fragmented or overly prescriptive rules. The principles recommend evaluating existing healthcare and consumer protection laws before introducing new AI-specific regulations and caution against broad prohibitions or mandatory pre-authorization schemes.
The ATA supports federal leadership in AI regulation to promote consistency across states and suggests that state laws generally exempt AI systems already regulated by federal agencies.
The updated guidance adds detail on validation and performance monitoring, urging developers and healthcare organizations to establish evidence of safety, reliability and clinical relevance for AI tools. The ATA argues that industry-led validation and monitoring can provide safeguards while reducing the need for rigid regulatory oversight.
Privacy and data security remain foundational elements of the framework. The principles state that AI systems should comply with established healthcare privacy standards while enabling responsible data use for model training and improvement.
Developers and deployers are encouraged to be transparent about data collection and processing practices, while regulatory approaches should protect proprietary information to support continued innovation.
Finally, the ATA addresses workforce implications, framing AI as a tool to augment clinicians rather than replace them. The principles emphasize reducing administrative burden, supporting new care models and addressing workforce shortages through collaboration among technology developers, providers and educators.
Nathan Eddy is a healthcare and technology freelancer based in Berlin.
Email the writer: nathaneddy@gmail.com
Healthcare IT News is a HIMSS Media publication
