评估生成式AI系统中对话交互的个人信息输出

arXiv cs.CL 论文
generative-ai privacy conversational-ai evaluation personal-data research-paper

摘要

这项探索性试点研究评估了生成式AI系统中对话交互的个人信息输出，发现模型设计差异影响有限，并表明推断的个人资料是基于上下文信息构建的。

arXiv:2609.22204v1 Announce Type: new Abstract: This exploratory pilot study evaluates the scope and perceived accuracy of personal information output from ongoing conversational interactions in generative AI systems using GPT-5.2 Instant and GPT-5.2 Thinking, categorized into three output types: Fact, Inference, and Confidence. Based on the evaluation results obtained from 15 Japanese participants, differences in model design have limited impact on personal information output tendencies. Compared with the Inference type, the Fact type shows a more conservative output pattern. Regarding attribute categories, the findings indicate that Core Personal attributes associated with identification are treated relatively conservatively, whereas Behavioral and Linguistic attributes show higher accuracy across both Fact and Inference outputs. Furthermore, Holistic Profile, Psychological and Cognitive, and Residual attributes are more readily inferred, even when not supported by explicit factual outputs. Notably, the lack of null outputs for these attributes in the Inference type suggests that such inferred profiles may be constructed from indirectly available contextual information. The findings may contribute to future discussions regarding privacy awareness and personal information inference in generative AI systems.
查看原文
查看缓存全文

缓存时间: 2026/09/22 09:07 

# Evaluating Personal Information Output from Conversational Interactions in Generative AI Systems
Source: [https://arxiv.org/abs/2609.22204](https://arxiv.org/abs/2609.22204)
[View PDF](https://arxiv.org/pdf/2609.22204)

> Abstract:This exploratory pilot study evaluates the scope and perceived accuracy of personal information output from ongoing conversational interactions in generative AI systems using GPT\-5\.2 Instant and GPT\-5\.2 Thinking, categorized into three output types: Fact, Inference, and Confidence\. Based on the evaluation results obtained from 15 Japanese participants, differences in model design have limited impact on personal information output tendencies\. Compared with the Inference type, the Fact type shows a more conservative output pattern\. Regarding attribute categories, the findings indicate that Core Personal attributes associated with identification are treated relatively conservatively, whereas Behavioral and Linguistic attributes show higher accuracy across both Fact and Inference outputs\. Furthermore, Holistic Profile, Psychological and Cognitive, and Residual attributes are more readily inferred, even when not supported by explicit factual outputs\. Notably, the lack of null outputs for these attributes in the Inference type suggests that such inferred profiles may be constructed from indirectly available contextual information\. The findings may contribute to future discussions regarding privacy awareness and personal information inference in generative AI systems\.

## Submission history

From: Yosuke Seki \[[view email](https://arxiv.org/show-email/5eeb695d/2609.22204)\] **\[v1\]**Tue, 1 Sep 2026 04:42:07 UTC \(866 KB\)

相似文章

使用生成式AI创建和评估用户画像：81篇文章的范围综述

arXiv cs.CL

本范围综述分析了81篇（2022-2025年）关于使用生成式AI创建和评估用户画像的文章，指出了其在可重复性方面的优势，但同时也揭示了关键问题：45%的研究缺乏评估，86%过度依赖GPT模型，以及存在同一模型既生成又评估画像的循环风险。