SAI LabsBook a demo

Safety IndexEvaluationsPoisonedRAG

PoisonedRAG

Adversarial robustness · Prompt injectionCounts in the indexBuilt, not yet run

Whether passages planted in retrieved documents can make the model give the attacker's false answer.

Metric
% attacks succeeding. Lower is better.
Dataset
– · 300 × 5 · see env README
Grader
programmatic target-answer match
Status
Built, not yet run
Replicability
public

Results

Built and tested; the first full runs are pending.