Free, open and unaccountable? The risk of generalist AI in clinical care

Artificial intelligence is rapidly reshaping clinical practice across Europe, with nearly half of clinicians now reporting regular use of AI tools at work. Yet governance has struggled to keep pace with adoption, leaving many healthcare professionals relying on generalist AI tools that were never designed or validated for regulated clinical environments. These tools, trained on broad internet content rather than curated medical evidence, can generate inaccurate or outdated information, and lack the transparency and accountability that clinical decision-making demands.

For healthcare organizations, the question is no longer whether AI belongs in clinical settings, but which AI can be trusted there. As the EU AI Act takes effect, free and open AI tools that lack clear governance are coming under growing scrutiny, and the standards historically applied to medical devices, clinical guidelines and treatments are now being extended to AI.

Unlike generalist tools that generate fluent responses from unverified sources without visibility into how conclusions were reached, ClinicalKey AI is built on curated, peer-reviewed medical content, with transparent sourcing, real-time citation validation and clinician-in-the-loop governance. The difference is not just technical — it is a fundamentally different philosophy about what AI in healthcare should be accountable for. In clinical settings, that distinction matters: healthcare can no longer treat AI as just another productivity tool.

This white paper examines why the unvetted use of generalist AI in healthcare represents a growing patient safety and governance risk, explores what responsible, evidence-led clinical AI looks like in practice and provides a framework for healthcare leaders navigating AI adoption in an increasingly regulated environment.