Preventing data leakage in AI agents for healthcare
ID: 81d76ae7-61f1-55c9-91a7-a36b3c0c6ebf
STIX ID: report--81d76ae7-61f1-55c9-91a7-a36b3c0c6ebf
Feed Name: Giskard
Threat Score
An AI transcription assistant auto-joined a hospital Zoom rounds meeting in September 2024 because a former physician remained on a recurring calendar invite; the agent recorded, transcribed, and emailed PHI to an unauthorized recipient. The report maps the failure to OWASP LLM02 (Sensitive Information Disclosure) and advises continuous automated red-teaming, LLM vulnerability scanning, and stricter tool-use and authorization checks to prevent similar data leaks.
Your team is not currently subscribed to this feed. You must subscribe to it in order to see this post.