MIT scientists investigate memorization risk in the age of clinical AI

New research nstrates how AI models can be tested to ensure they don’t cause harm by revealing anonymized patient health data.