SRPG: Semantically Reconstructed Privacy Guard for Zero-Trust Privacy in Educational Multi-Agent Systems
On the map: LLM agents and tool use → Agent threats, guardrails and governance
Abstract (arXiv)
Multi-Agent Systems (MAS) with large language models (LLMs) enable personalized education but risk leaking minors personally identifiable information (PII) via unstructured dialogue. Existing privacy methods struggle to balance security and utility: role-based access control fails on unstructured text, while naive masking destroys pedagogical context. We propose SRPG, a privacy guard for educational MAS, using a Dual-Stream Reconstruction Mechanism: a strict sanitization stream ensures zero PII leakage, and a context reconstruction stream (LLM driven) recovers mathematical logic. This decouples instructional content from private data, preserving teaching efficacy. Tests on MathDial show SRPG works across models; with GPT-4o, it achieves 0.0000 Attack Success Rate (ASR) (zero leakage) and 0.8267 Exact Match, far outperforming the zero trust Pure LLM baseline (0.2138). SRPG effectively protects minors privacy without sacrificing mathematical instructional quality.
Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Agent threats, guardrails and governance, one of 16 directions in LLM agents and tool use. Builds on 7 of its 18 references that are in the map. First posted on arXiv 3 Dec 2025.
Builds on
7 of 18 references.
Closest ancestors · 1
Same direction, newest first
Other references · 6
- 2020Language Models are Few-Shot Learnerscited by 32,419
- 2023Generative Agents: Interactive Simulacra of Human Behaviorcited by 3,568
- 2020Extracting Training Data from Large Language Modelscited by 2,189
- 2023MetaGPT: Meta Programming for A Multi-Agent Collaborative Frameworkcited by 1,808
- 2022What Does it Mean for a Language Model to Preserve Privacy?cited by 195
- 2019Privacy- and Utility-Preserving Textual Analysis via Calibrated Multivariate Perturbationscited by 119
Built on by
Not cited yet.
Similar, not cited
Close in wording, no citation either way
- 2026Privacy-Preserving RAG via Multi-Agent Semantic Rewriting: Achieving Confidentiality Without Compromising Contextual Fidelity
- 2026PrivAct: Internalizing Contextual Privacy Preservation via Multi-Agent Preference Training
- 20251-2-3 Check: Enhancing Contextual Privacy in LLM via Multi-Agent Reasoning
- 2025Privacy-R1: Privacy-Aware Multi-LLM Agent Collaboration via Reinforcement Learning
- 2026It Takes Two: Complementary Self-Distillation for Contextual Integrity in LLMs
- 2024Preserving Privacy in Large Language Models: A Survey on Current Threats and Solutions
- 2026Contextualized Privacy Defense for LLM Agents
- 2025PrivacyBench: A Conversational Benchmark for Evaluating Privacy in Personalized AI