#memorizationrisk — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #memorizationrisk, aggregated by home.social.
-
A new MIT study reveals that large language models can unintentionally memorize details from de‑identified patient notes, exposing a privacy gap in clinical AI. Researchers led by Arash Tonekaboni show why stricter safeguards are needed. Read on to see the implications. #MITAI #ClinicalAI #DeIdentifiedData #MemorizationRisk
🔗 https://aidailypost.com/news/mit-study-probes-memorization-risk-clinical-ai-deidentified-data