{"data":{"id":"ae7da0b5-1b6a-48b1-9c3f-6dd21b565faa","title":"COMPASS: Membership Inference Attack Against Vision Large Language Models via Adaptive Noise Calibration","summary":"COMPASS is an adaptive gray-box membership inference attack against vision large language models (VLLMs) that calibrates both perturbation magnitude and membership-score direction using a known non-member reference set. The authors find that a low-noise regime often separates members from non-members better than large perturbations, with the score direction reversing in some model-dataset pairs. Tested on LLaVA v1.5-7B, MiniGPT-4 and HuluMed-32B, COMPASS reaches AUCs up to 0.88 on general-domain data and 0.64 on distribution-matched medical data, where the strongest baseline stays at chance.","solution":"N/A -- no mitigation discussed in source.","labels":["security","privacy","research"],"sourceUrl":"https://arxiv.org/abs/2610.17332v1","publishedAt":"2026-10-12T19:41:52.000Z","cveId":null,"cweIds":null,"cvssScore":null,"cvssSeverity":null,"severity":"info","attackType":["membership_inference"],"issueType":"research","affectedPackages":null,"affectedPackageNames":null,"affectedPackageRefs":null,"affectedVendors":[],"affectedVendorsRaw":["LLaVA v1.5-7B","MiniGPT-4","HuluMed-32B"],"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":"2026-10-12T19:41:52.000Z","capecIds":null,"crossRefCount":0,"attackSophistication":"advanced","impactType":["confidentiality"],"aiComponentTargeted":"model","llmSpecific":true,"classifierConfidence":0.93,"researchCategory":"preprint","atlasIds":["AML.T0024.000"]}}