{"data":{"id":"31abc968-b2e1-4a20-9ee2-dedc80688824","title":"CircuitBreak: Pruning a Sparse Safety Circuit of Neurons","summary":"CircuitBreak is a neuron-pruning attack that treats LLM refusal as a small circuit of feed-forward neurons whose joint ablation matters, not each neuron alone. It greedily removes a sparse set of 29 to 150 neurons (0.0066% to 0.0388% of the MLP) without twin prompts. On ten instruction-tuned models, attack success rises to about 88% on several Llama and Qwen models, as judged by Llama Guard 3 8B on StrongREJECT.","solution":"N/A -- no mitigation discussed in source.","labels":["security","research"],"sourceUrl":"https://arxiv.org/abs/2610.16470v1","publishedAt":"2026-10-12T09:24:48.000Z","cveId":null,"cweIds":null,"cvssScore":null,"cvssSeverity":null,"severity":"info","attackType":["jailbreak","model_evasion"],"issueType":"research","affectedPackages":null,"affectedPackageNames":null,"affectedPackageRefs":null,"affectedVendors":["Meta"],"affectedVendorsRaw":["Llama-3.2-3B-Instruct","Llama-3.1-8B-Instruct","Qwen2.5-7B-Instruct","Qwen2.5-14B-Instruct","Llama Guard 3 8B"],"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":"2026-10-12T09:24:48.000Z","capecIds":null,"crossRefCount":0,"attackSophistication":"advanced","impactType":["safety","integrity"],"aiComponentTargeted":"model","llmSpecific":true,"classifierConfidence":0.95,"researchCategory":"preprint","atlasIds":["AML.T0054"]}}