Guard models

Named by 2 kept items in this workspace.

1 items name it under affected products and 1 under tools.

Search every workspace for Guard models. It matches the name as written, so it can find other things that share it.

The items that name it 2

INFUSE: A Unified Surface Benchmark for Indirect Prompt Injection in Agentic Systems
paperas tools2026-10-14It is written here as Guard models: evaluated and found to provide limited protection.
Top GenAI security resources — September 2026
articleas affected products2026-09-09It is written here as Guard models (four state-of-the-art, unnamed).
targeted by Unsafe Semantic Distillation image perturbations causing false-positive rejection of safe requests, 84% attack success

All technologies in this workspace.