IMPersona: Evaluating Individual Level LM Impersonation
This addresses privacy and security risks from personalized language models, though it is incremental as it builds on existing fine-tuning and retrieval methods.
The paper tackled the problem of evaluating how well language models can impersonate specific individuals' writing style and personal knowledge, finding that fine-tuned models with memory integration were misidentified as human in 44.44% of blind conversations, compared to 25.00% for prompting-based approaches.
As language models achieve increasingly human-like capabilities in conversational text generation, a critical question emerges: to what extent can these systems simulate the characteristics of specific individuals? To evaluate this, we introduce IMPersona, a framework for evaluating LMs at impersonating specific individuals' writing style and personal knowledge. Using supervised fine-tuning and a hierarchical memory-inspired retrieval system, we demonstrate that even modestly sized open-source models, such as Llama-3.1-8B-Instruct, can achieve impersonation abilities at concerning levels. In blind conversation experiments, participants (mis)identified our fine-tuned models with memory integration as human in 44.44% of interactions, compared to just 25.00% for the best prompting-based approach. We analyze these results to propose detection methods and defense strategies against such impersonation attempts. Our findings raise important questions about both the potential applications and risks of personalized language models, particularly regarding privacy, security, and the ethical deployment of such technologies in real-world contexts.