{"data":{"id":"097f8e50-d9f5-47f8-a0dd-ee7c6e1a2a4d","title":"Policy-Verified Self-Training: Learning Agent Compliance from Verifiable Policy Rewards","summary":"Runtime policy verifiers usually discard their enforcement decisions, but this paper asks whether those verdicts can train the agent instead. The authors propose Policy-Verified Self-Training (PVST), where a deterministic gate compiled from a written policy scores the agent's own trajectories, reinforcing compliant ones and using violations as negatives for preference optimization. On held-out AgentDojo tasks under prompt injection, training Qwen3-8B cut violating trajectories from 37.2% to 10.3% and successful attacks from 12.4% to 0.5%, though utility fell on travel tasks.","solution":"N/A -- no mitigation discussed in source.","labels":["security","research"],"sourceUrl":"https://arxiv.org/abs/2610.19372v1","publishedAt":"2026-10-14T01:16:15.000Z","cveId":null,"cweIds":null,"cvssScore":null,"cvssSeverity":null,"severity":"info","attackType":["prompt_injection"],"issueType":"research","affectedPackages":null,"affectedPackageNames":null,"affectedPackageRefs":null,"affectedVendors":[],"affectedVendorsRaw":["Qwen3-8B","AgentDojo"],"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":"2026-10-14T01:16:15.000Z","capecIds":null,"crossRefCount":0,"attackSophistication":"moderate","impactType":["integrity","safety"],"aiComponentTargeted":"agent","llmSpecific":true,"classifierConfidence":0.95,"researchCategory":"preprint","atlasIds":["AML.T0051"]}}