{"data":{"id":"4479cefe-123b-4057-bda2-1cc168673873","title":"Heads of Security: Learned Attention Gates Improve Prompt Injection Robustness","summary":"Researchers introduce Heads of Security (HoS), a prompt injection defense that uses a small set of learned per-head attention gates to separate trusted instructions from untrusted data. Evaluated on Llama-3.3-70B-Instruct and GPT-OSS-20B, HoS cut the black-box RL-Hammer attack success rate to 10%, versus 97% on GPT-OSS-20B and 85% on SecAlign. It also preserved base-model utility on standard benchmarks, where SecAlign and ASIDE degraded in few-shot settings.","solution":"N/A -- no mitigation discussed in source.","labels":["security","research"],"sourceUrl":"https://arxiv.org/abs/2610.37657v1","publishedAt":"2026-10-27T17:58:52.000Z","cveId":null,"cweIds":null,"cvssScore":null,"cvssSeverity":null,"severity":"info","attackType":["prompt_injection"],"issueType":"research","affectedPackages":null,"affectedPackageNames":null,"affectedPackageRefs":null,"affectedVendors":[],"affectedVendorsRaw":["Llama-3.3-70B-Instruct","GPT-OSS-20B","SecAlign++","ASIDE"],"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":"2026-10-27T17:58:52.000Z","capecIds":null,"crossRefCount":0,"attackSophistication":"advanced","impactType":["integrity","safety"],"aiComponentTargeted":"model","llmSpecific":true,"classifierConfidence":0.97,"researchCategory":"preprint","atlasIds":["AML.T0051"]}}