{"data":{"id":"52068578-818e-4893-9be1-6a6fd0e36ec5","title":"SafeStart: Mitigating the Overlooked Safety Cost of Active Expert Reduction in MoE LLM Serving","summary":"This research shows that serving methods which run fewer experts per token in Mixture-of-Experts language models can weaken refusal of harmful requests. Under a jailbreak, harmful compliance rose by 18.4 points on Qwen1.5-MoE, 9.5 on OLMoE and 6.6 on DeepSeek-V2-Lite. The authors introduce SafeStart, which runs all experts for the prompt and the first eight generated tokens, and keeps at least 85% of the offline decode speedup at batch sizes 8 and 32.","solution":"SafeStart runs all experts for the prompt and the first eight generated tokens, and the reduced set of experts afterwards, through a fused kernel in vLLM. The source also advises serving systems to test refusal, not only accuracy, before enabling expert-reduction methods.","labels":["research","safety"],"sourceUrl":"https://arxiv.org/abs/2611.06147v1","publishedAt":"2026-11-04T08:31:01.000Z","cveId":null,"cweIds":null,"cvssScore":null,"cvssSeverity":null,"severity":"info","attackType":["jailbreak"],"issueType":"research","affectedPackages":null,"affectedPackageNames":null,"affectedPackageRefs":null,"affectedVendors":[],"affectedVendorsRaw":["Qwen1.5-MoE","OLMoE","DeepSeek-V2-Lite","vLLM","MoE LLMs"],"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":"2026-11-04T08:31:01.000Z","fixReleasedAt":null,"daysToFix":null,"fixCheckedAt":null,"capecIds":null,"crossRefCount":0,"attackSophistication":"moderate","impactType":["safety"],"aiComponentTargeted":"inference","llmSpecific":true,"classifierConfidence":0.93,"researchCategory":"preprint","atlasIds":["AML.T0054"]}}