PAAC: Privacy-Aware Agentic Device-Cloud Collaboration
Liangqi Yuan, Wenzhi Fang, Shiqiang Wang, Christopher G. Brinton

TL;DR
PAAC introduces a privacy-aware framework for device-cloud LLM agents, balancing reasoning capabilities with privacy preservation by aligning decomposition with privacy mechanisms, leading to significant improvements in privacy and accuracy.
Contribution
It presents a novel agentic framework that uses role specialization as a privacy mechanism, improving privacy-accuracy trade-offs in device-cloud LLM collaboration.
Findings
PAAC outperforms state-of-the-art baselines in privacy and accuracy on three benchmarks.
PAAC improves average accuracy by 15-36% and reduces leakage by 2-6 times.
Consistent improvements observed across 17 benchmarks in diverse domains.
Abstract
Large language model (LLM) agents face a structural tension: cloud agents provide strong reasoning but expose user data, while on-device agents preserve privacy at the cost of overall capability. Existing device-cloud designs treat this boundary as a compute split rather than a trust boundary suited to agentic workloads, and existing sanitizers force a choice between policy flexibility and the structural fidelity tool calls require. In this work, we develop PAAC, a privacy-aware agentic framework that aligns planner--executor decomposition with the device-cloud boundary so that role specialization itself becomes the privacy mechanism. The cloud agent reasons over typed placeholder tokens that preserve each sensitive value's reasoning role while discarding its content, while the on-device agent identifies sensitive spans and distills each step's execution outcome into compact key…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
