Research · curated 9 Aug 2026

Prompt injection attacks on vision-language models for surgical decision support

Coverage timeline

9 Aug 2026nih.gov

Single-source research — first reported, latest, and curated coincide.

Why it matters

Prompt injection against medical VLMs shows how safety-critical clinical AI systems can be manipulated, extending the injection threat surface into surgical decision support.

A PubMed-indexed study examines prompt injection attacks against vision-language models (VLMs) used for video-based surgical decision support, probing how such models can be manipulated in high-stakes clinical settings. The abstract notes VLMs' promise for understanding complex temporospatial surgical video but investigates their vulnerability to injected prompts.