Hugging Face has introduced MosaicLeaks to address a privacy risk in deep research agents that combine private local documents with external web tools. When agents execute external queries, an observer can reassemble fragments of information to deduce sensitive enterprise details, a phenomenon known as the mosaic effect. MosaicLeaks measures this leakage across intent, answer, and full-information categories using a dataset of 1,001 multi-hop research chains over local enterprise documents and a controlled web corpus. According to testing, when agents are trained solely for task performance, privacy leakage worsens because models learn to pack more context into their web queries. To resolve this tension, Hugging Face proposes Privacy-Aware Deep Research, a reinforcement learning method that utilizes situational task rewards and a learned privacy reward. This training approach raises strict chain success from 48.7% to 58.7% while reducing answer and full-information leakage from 34.0% to 9.9%. The method allows agents to continue issuing necessary web queries while dropping revealing details and specific metrics from the query text.
MosaicLeaks: Can your research agent keep a secret?
Deep research agents that combine private local documents with external web tools pose significant privacy risks through the mosaic effect, where cumulative external queries leak sensitive enterprise data. Hugging Face's MosaicLeaks proposes a new deep-research task and a reinforcement learning training method called Privacy-Aware Deep Research to mitigate this issue.
Why it matters
As enterprise AI agents increasingly rely on both internal proprietary data and external web retrieval tools, safeguarding private information against leakage via cumulative query logs is a critical security challenge for the industry.
Topic dossiers Hugging Face