Health Affairs August 26, 2025
Artificial intelligence (AI) is reshaping medicine, driving tools that interpret imaging, generate clinical notes, and assist in diagnosis and triage. These advances rely on vast stores of patient data—and on the public’s willingness to share it. Yet, as AI systems grow more capable, so too does their capacity to remember. The promise of AI thus carries a paradox: To sustain trust, machines may need to learn how to forget.
Machine unlearning—the targeted removal of data from a trained AI model—has emerged as a potential safeguard against AI models retaining sensitive data even after it’s meant to be erased.
The concept aligns with global moves toward stronger data privacy protections, including the European Union’s “right to be forgotten,” which gives individuals...
Today's Sponsors
Today's Sponsor