🧠 AI Ed Wiki

📄 Research Article

When Agents Learn to Be You: Benchmarking Privacy Leakage, Impersonation Risk, and Defenses in Persona Skills

● high · 2026-08-05privacy, agentic-ai, student-ai-interaction, bias-mitigation, personalized-learning, benchmark, rag

When Agents Learn to Be You: Benchmarking Privacy Leakage, Impersonation Risk, and Defenses in Persona Skills — Introduces AntiSkillBench with 7,500 persona-grounded dialogue traces from 50 behaviorally rich profiles. Evaluates skill-level privacy leakage, agent-level attribute disclosure, and behavioral impersonation across three skill-distillation strategies... Privacy Agentic AI Student Experience Bias Mitigation Personalized Learning benchmark

Introduces AntiSkillBench with 7,500 persona-grounded dialogue traces from 50 behaviorally rich profiles. Evaluates skill-level privacy leakage, agent-level attribute disclosure, and behavioral impersonation across three skill-distillation strategies. Experiments across three frontier agents show persona-skill risks persist across agent backbones and distillation protocols, extending from explicit attributes to communication styles and personality traits. Existing defenses exhibit limited and distillation-dependent effectiveness, failing to generalize across risk and distillation strategies.

Abstract

Persona skills distill personal interaction histories into portable and executable artifacts for downstream agents. While enabling flexible personalization, this process concentrates fragmented personal signals, amplifies their impact through reuse, and challenges defenses designed for individual records or retrieval-based memory. To systematically investigate the safety of the persona-skill pipeline, we introduce AntiSkillBench, an end-to-end benchmark for evaluating risks and defenses across the persona-skill pipeline.

Connected Concepts

  • Privacy
  • Agentic AI
  • Student Experience
  • Bias Mitigation
  • Personalized Learning

    • Connected Articles

  • Learning Engagement Assistant Lea — Learning Engagement Assistant (LEA): Cross-Course Scalability and Classroom Evaluation of an Agentic AI Tutoring System
  • Cstutorbench Slm Tutors — CSTutorBench: Benchmarking Small Language Models as Tutors for Block-Based Programming
  • MOOC To Maic — From MOOC to MAIC: Reshaping Online Teaching and Learning through LLM-driven Agents
  • Trust Reliance AI Education 2026 — Trust and Reliance on AI in Education: AI Literacy and Need for Cognition as Moderators
  • Deeptutor — DeepTutor: Towards Agentic Personalized Tutoring
  • Jeon Isd Agent Bench 2026 — ISD Agent Benchmark

    • Citation

    Yongli Xiang, Zhifang Zhang, Bojun Yang, Ziming Hong, Lei Feng, Miao Xu, & Tongliang Liu (2026). When Agents Learn to Be You: Benchmarking Privacy Leakage, Impersonation Risk, and Defenses in Persona Skills. arXiv:2608.03700. arXiv:2608.03700 [cs.CR].