Abstract Generative agents are a novel AI agent architecture developed in 2023 using LLMs that can generate believable human behaviors. They are of significant importance in social simulation and human-computer interaction. However, there is a lack of research on the privacy and security issues that may arise from its subsequent application in society. Based on the Generative Agent framework, this paper reintroduces a personal information system, and a shared database, reconstructing the memory and reflection systems to realize a Lightweight Generative Agent framework (LGA). To fill the research gap in the field of privacy for Generative Agents, this paper simulates three scenarios with five agents to explore potential privacy and security issues. Based on observed privacy leaks, three defensive strategies are proposed: behavior guideline regulation, Multi-Agents cooperation, and step-back prompts. These strategies have reduced privacy leaks by more than 50.00% on average. For some privacy leaks with special simulation scenarios, it has even reduced privacy leaks by more than 80.00%. Additionally, through supplementary experiments, we demonstrate that LGA achieves better cost-effectiveness compared to the original Generative Agent framework. The analysis of privacy and security issues provides a reference for subsequent research on the safety of generative agents.
Guo et al. (Sun,) studied this question.