{"data":{"id":"7f470492-91bb-4bca-ac04-19eb1cb9b6ff","title":"PIPETTE: Prompt Injection Purple-Teaming for Robust LLM Agents","summary":"Researchers show that RL-based red-teaming methods, RL-Hammer and PISmith, suffer from mode collapse: they repeatedly find similar indirect prompt injection strategies, so high attack success rates during red-teaming do not produce robustness gains in later adversarial training. They introduce PIPETTE, a purple-teaming framework that uses a dictionary learning-based reward for broader attack coverage and an on-policy adversarial training pipeline. On AgentDojo and ASB, PIPETTE yields defenders with near-zero attack success rates on static attacks that withstand strong adaptive attacks, with benign task utility close to the undefended baseline.","solution":"PIPETTE, a unified prompt injection purple-teaming framework that converts red-teaming discoveries into defender robustness improvements, using a dictionary learning-based reward and an on-policy adversarial training pipeline tailored to IPI defense.","labels":["security","research"],"sourceUrl":"https://arxiv.org/abs/2611.24362v1","publishedAt":"2026-11-18T12:56:06.000Z","cveId":null,"cweIds":null,"cvssScore":null,"cvssSeverity":null,"severity":"info","attackType":["prompt_injection"],"issueType":"research","affectedPackages":null,"affectedPackageNames":null,"affectedPackageRefs":null,"affectedVendors":[],"affectedVendorsRaw":["LLM agents","RL-Hammer","PISmith","AgentDojo","ASB"],"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":"2026-11-18T12:56:06.000Z","fixReleasedAt":null,"daysToFix":null,"fixCheckedAt":null,"capecIds":null,"crossRefCount":0,"attackSophistication":"advanced","impactType":["integrity","safety"],"aiComponentTargeted":"agent","llmSpecific":true,"classifierConfidence":0.95,"researchCategory":"preprint","atlasIds":["AML.T0051"]}}