Label-Budgeted Attribute Inference from Protected Features in Edge-Cloud Split Multimodal LLMs

Summary

Researchers show that in edge-cloud split inference of multimodal LLMs, an honest-but-curious server can infer age, gender and race from the intermediate features it receives, even when raw images stay local. They propose a label-budgeted recovery curve to measure feature-level privacy, finding that Safe-LLaVA, which filters model responses, still lets a server recover 75-85% of unprotected leakage from transmitted features with 20 labeled samples per class.

Topics

• Multimodal models

Related items

• InfoBeyond membership inference: Distributional leakage in diffusion modelsSimilar attack · Elsevier Security Journals
• LowTRACE: Data-Free Text Reconstruction Attacks against Approximate Unlearning in LLMsSimilar attack · Arxiv (cs.CR + cs.CL + cs.LG)
• InfoAfterDeleteBench: Auditing Post-Deletion Information Persistence in LLM AgentsSimilar attack · Arxiv (cs.CR + cs.CL + cs.LG)
• InfoEnterprise Agentic RAG with Pre-Search Role-Based Access Control, Multi-Layered Guardrails, and Operational TelemetrySimilar attack · OpenAlex (peer-reviewed AI security)
• LowAgent Mosaic: Recognizing Confidentiality Leaks in Agentic LLM SystemsSimilar attack · Arxiv (cs.CR + cs.CL + cs.LG)