{"data":{"id":"da6140b7-f7c6-4112-a7af-75e472dae067","title":"Identify, Classify, Act: Per-Instruction Controlled Compliance for LLMs over External Data","summary":"This paper proposes EDIC, a decomposed reward framework that trains LLMs to classify each instruction embedded in external data as beneficial or harmful, then follow the former and reject the latter. A generative judge, initialized by SFT and refined with GRPO, attributes compliance per instruction to produce separate rewards. Across two backbones and five benchmarks, EDIC improves harmful-rejection by up to 7.7% and Arena-Hard-Auto utility by 12.5% on Qwen3-4B over the strongest prior defense.","solution":"N/A -- no mitigation discussed in source.","labels":["research","security"],"sourceUrl":"https://arxiv.org/abs/2610.22874v1","publishedAt":"2026-10-16T03:24:08.000Z","cveId":null,"cweIds":null,"cvssScore":null,"cvssSeverity":null,"severity":"info","attackType":["prompt_injection"],"issueType":"research","affectedPackages":null,"affectedPackageNames":null,"affectedPackageRefs":null,"affectedVendors":[],"affectedVendorsRaw":["Qwen3-4B"],"classifierModel":"claude-haiku-5-5","classifierPromptVersion":"v4","summaryPromptVersion":"v2","headline":null,"headlinePromptVersion":null,"cvssVector":null,"attackVector":null,"attackComplexity":null,"privilegesRequired":null,"userInteraction":null,"exploitMaturity":null,"epssScore":null,"epssCheckedAt":null,"kevDateAdded":null,"advisoryAliases":null,"affectedPackagesSource":null,"affectedPackagesCheckedAt":null,"patchAvailable":null,"disclosureDate":"2026-10-16T03:24:08.000Z","capecIds":null,"crossRefCount":0,"attackSophistication":"moderate","impactType":["integrity","safety"],"aiComponentTargeted":"agent","llmSpecific":true,"classifierConfidence":0.95,"researchCategory":"preprint","atlasIds":["AML.T0051"]}}