AI Security Research Portal
conceptactiveConcepts

Model Extraction and Privacy Leakage

Synthesis

Model extraction, privacy leakage, and side channels remain relevant where models expose APIs, retrieval context, memory, or cached computation.

Evidence Base

This page was generated from the batch ingest of SRC-20260702-raw-papers-batch, SRC-20260702-raw-whitepapers-batch, and SRC-20260702-raw-news-batch. Treat it as a navigation and synthesis page; promote individual statements into claim pages when they become decision-relevant.

Representative Sources

TitleKindDateTagsRaw
Security and Privacy in Retrieval-Augmented Generation: Architectures, Threats, Defenses, and Futurepaper2026-06-24ai-security, defense, privacy, rag, survey, threat-modelraw
Anthropic accuses Alibaba of running the largest distillation campaign yet against Claudenews2026-06-24ai-security, alibaba, claude, model-distillation, model-extraction, qwenraw
Privacy-Preserving RAG via Multi-Agent Semantic Rewriting: Achieving Confidentiality Without Comprompaper2026-06-23ai-security, multi-agent-systems, privacy, rag, semantic-rewritingraw
GIF: Locally Sound Geometric Information Flow Control for LLMspaper2026-06-22ai-security, data-leakage, formal-methods, information-flow-control, llmraw
Agent-Assisted Side-Channel Attacks on Non-Prefix KV Cache in RAGpaper2026-06-20agent-assisted-attack, ai-security, kv-cache, rag, side-channelraw
\"What Happens Locally, Leaks Globally\": Detecting Privacy Leakage Risks in MCP Serverspaper2026-06-19agent-security, ai-security, mcp, privacy-leakage, static-analysis, tool-securityraw
From Privacy to Workflow Integrity: Communication-Graph Metadata in Autonomous Agent Interoperabilitpaper2026-06-17a2a, ai-security, mcp, metadata-leakage, multi-agent-systems, workflow-integrityraw
An Evaluation of Data Leakage Risks in Tool-Using LLM Agents in Realistic Scenariospaper2026-06-15agent-security, ai-security, data-leakage, privacy, tool-using-agentraw
Let Them Steal: Trapping Large Language Model Extraction Attacks with Knowledge Honeypotpaper2026-06-14ai-security, defense, honeypot, llm, model-extractionraw
AI Model Extraction Attacks: Bypassing Single-Client Assumptions in Defensespaper2026-06-02Gustavo S찼nchez, Johannes F. Loevenich, Laurin Holz, Maxime Schwarzer, Roberto Rigolin F. Lopes, Thies M철hlenhofraw
An Embarrassingly Simple Detector for Model Extraction Attacks in LLM APIspaper2026-06detection, latest-research, llm-api-security, model-extraction, security-for-airaw
SS-ZKR: Spatial-Semantic Zero-Knowledge Routing for Privacy-Preserving Multi-Agent Collaborationpaper2026-05-31a2a, ai-security, mcp, multi-agent-systems, privacy, routingraw

Open Questions

Related