When Agents Learn to Be You: Benchmarking Privacy Leakage, Impersonation Risk, and Defenses in Persona Skills
Yongli Xiang, Zhifang Zhang, Bojun Yang, Ziming Hong, Lei Feng, Miao Xu, Tongliang Liu
cs.CR, cs.CL, cs.CY
Submitted: 2026-08-04
Comments: Project page: https://yonglixiang.github.io/AntiSkillBench
Project page: https://yonglixiang.github.io/AntiSkillBench
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- MemPrivacy: Privacy-Preserving Personalized Memory Management for Edge-Cloud Agents
- Poise: Position-Aware One-Instruction Skill Injection for Silent Execution on LLM Agents
- ToolPrivacyBench: Benchmarking Purpose-Bound Privacy in Tool-Using LLM Agents
- Delving into the Temporal Challenges of Unified Video Protection Against Image-to-Video and Fine-Tuning-based Customization
- LongLaMP: A Benchmark for Personalized Long-form Text Generation
- A Survey of Personalized Large Language Models: Progress and Future Directions
- MemoChat: Tuning LLMs to Use Memos for Consistent Long-Range Open-Domain Conversation
- PrivacyBench: A Conversational Benchmark for Evaluating Privacy in Personalized AI
- IMPersona: Evaluating Individual Level LM Impersonation
- Voyager: An Open-Ended Embodied Agent with Large Language Models
- OpenCharacter: Training Customizable Role-Playing LLMs with Large-Scale Synthetic Personas
- DAMR: Efficient and Adaptive Context-Aware Knowledge Graph Question Answering with LLM-Guided MCTS
- DisRFM: Polar Riemannian Flow Matching for Structure-Preserving Graph Domain Adaptation
- Understanding the Role of User Profile in the Personalization of Large Language Models
- When Safety Collides: Resolving Multi-Category Harmful Conflicts in Text-to-Image Diffusion via Adaptive Safety Guidance
- Agent Skills for Large Language Models: Architecture, Acquisition, Security, and the Path Forward
- Inferential Privacy Leakage in Anonymized Conversational AI Logs
- PrivacyPeek: Auditing What LLM-Based Agents Acquire, Not Just What They Say
- BERTScore: Evaluating Text Generation with BERT
- Personalization of Large Language Models: A Survey
Related papers
- SoK: AI-Augmented Binary Reversing
- Relaxed Sender Anonymity for CBDC Interbank Settlement: A Zero-Knowledge Approach on Permissioned EVM
- Calibration-Family Overfit: Why Trusted Sabotage Monitors Don't Transfer Across Lineages
- Efficient Fuzzy PSI under One-Sided Assumptions
- Sealing the Audit-Runtime Gap for LLM Skills
- Token Composition: A Graph Based on EVM Logs