Regulatory Approval Is Not Enough: Gaps in Trustworthy AI Reporting in FDA-Cleared Medical Devices

Regulatory approval alone doesn't ensure trustworthiness: Nearly one-quarter (24.7%) of FDA-cleared AI medical devices provided no documented evidence of trustworthy AI practices, and none of the 519 devices analyzed demonstrated all six key trustworthiness principles, indicating that FDA clearance does not guarantee the transparency needed for independent assessment of AI safety and reliability.

Critical gaps exist in explaining how AI makes decisions: Explainability was documented in only 3.5% of devices and Traceability in just 8.3%, meaning most cleared AI systems lack publicly available information about how they reach clinical decisions or how their outputs can be traced back to specific inputs—essential for clinicians and patients to understand and trust these tools.

Robustness receives attention, but other dimensions lag: While 57.6% of devices reported evidence of robustness (testing across different conditions), the uneven reporting across the six trustworthiness principles suggests that manufacturers and regulators prioritize some aspects of AI safety while neglecting others like fairness and usability.

Reporting standards haven't improved over time: The year a device received FDA clearance did not predict better documentation of trustworthy AI principles, suggesting that evolving regulatory expectations have not yet translated into more transparent reporting practices by manufacturers.

Public documentation is insufficient for independent oversight: The limited and inconsistent evidence in publicly available FDA summaries means clinicians, patients, and independent researchers cannot adequately evaluate whether cleared AI devices meet trustworthiness standards before deployment in clinical settings.