{"arxiv_id":"2610.06401","title":"RAISED: Self-Distillation for Robustness to Prompt Injection in LLM Agents","abstract":"Tool-using language-model agents are vulnerable to indirect prompt injection because they must act on untrusted external content. Existing training-time defenses can reduce attack success rates, but often at the cost of general capabilities. We show that training-based defenses induce substantial drift in the model's output distribution, altering its behavior even in benign settings and providing a potential mechanism for utility degradation. We further identify a failure mode of these defenses: On benign tool-use tasks, the model refrains from a step needed to finish an authorized task, particularly when that step is indicated by a tool output. To address these limitations, we introduce RAISED (Robust Attack Invariance through Self-Distillation), a training framework that combines self-generation and self-distillation. The model first generates its own tool-use scenarios, with an emphasis on cases where task completion requires acting on legitimate guidance from tool outputs. Then, through self-distillation, the student is trained to match the teacher's clean-context behavior on both clean and injected variants of the same trajectory. RAISED substantially reduces the attack success rate of prompt injections in tool responses while, unlike prior training-based defenses, preserving utility on both agentic and general-purpose benchmarks.","primary_cat":"cs.CR","cross_cats":["cs.AI","cs.CL","cs.LG"],"submitted_at":"2026-10-05T14:21:30+00:00","latest_version":2,"source":"arxiv","source_label":"arXiv","source_url":null,"has_pdf":true,"has_source":false,"verdict":null,"confidence":null,"one_line_summary":null,"novelty_score":null,"correctness_risk":null,"formal_verification":null,"upvotes":0,"downvotes":0,"pith_score":0.0,"inbound_count":1,"_links":{"html":"https://pith.science/paper/2610.06401","verdict":"/api/v1/verdicts/2610.06401"}}