Erase the Injected Region: Multimodal Prompt-Injection Defense with On-Policy Self-Distillation

The paper introduces ERASE, a multimodal defense against indirect prompt injection that erases malicious instructions directly within the input media. By replacing the injected region with modality‑appropriate blank content and using on‑policy self‑distillation, ERASE trains a student model to learn from a frozen base model’s erased view, providing token‑level feedback during its own rollouts. Experiments on image, audio, and video tasks show near‑zero attack success while preserving task performance close to clean‑input levels, with significant improvements over prior methods such as ARGUS.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv AI.

• efficiency
• multimodal
• benchmarks