How Hospitals’ Use of GenAI is Putting Patients at Risk Without Realizing It

The article provides a framework for how healthcare organizations can adopt Generative AI/LLMs across four approaches—public models, private instances, embedded SDK/API integrations, and bespoke models—highlighting associated cyber and compliance risks (e.g., HIPAA/PHI exposure, bias, hallucinations, auditability, and vendor dependencies). It references OWASP’s GenAI Top 10 and industry observations of increasing attacks on GenAI apps to underscore the need for rigorous governance, validation by clinicians, and careful, controlled deployment. The guidance emphasizes selecting the implementation that balances risk and benefit while instituting strong security and compliance controls throughout the AI lifecycle.