Extracting Training Dialogue Data From Large Language Model-Based Task Bots

Summary

Researchers study whether LLM-based task-oriented dialogue systems memorize their training dialogue data, which can include phone numbers and full travel schedules. They evaluate existing training data extraction attacks, find that task-oriented dialogue traits make them ineffective, and propose new attack techniques for response sampling and membership inference. Their method extracts thousands of training dialogue-state labels with best-case precision above 70%, and the paper analyzes the factors that influence this memorization.

Topics

• Training data and privacy

Related items

• MediumCVE-2026-121938: langchain4j-aideepin before 3.31.0 contains an insecure direct object reference vulnerability that allows authenticated…Similar attack · NVD/CVE Database
• MediumGoogle: LLMjacking attacks rack up $1 million in cloud AI chargesSimilar attack · BleepingComputer
• InfoResidual Attribute Leakage Is an Interface: Auditing Privacy Evaluation After Text De-identificationSimilar attack · Arxiv (cs.CR + cs.CL + cs.LG)
• LowExtracting System Prompt of LLM Services in Complex (Agentic) Settings: Hierarchical Optimization and Reliable Evaluation MetricsSimilar attack · Arxiv (cs.CR + cs.CL + cs.LG)
• InfoMCP Tool Poisoning in Agentic AI Frameworks: A Cross-Framework Measurement StudySimilar attack · Arxiv (cs.CR + cs.CL + cs.LG)