arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2403.18249cs.CLcs.SI
收藏

探索LLM生成假新闻的欺骗能力:一项关于真实世界检测挑战的研究

Exploring the Deceptive Power of LLM-Generated Fake News: A Study of Real-World Detection Challenges

Virginia Tech(弗吉尼亚理工大学);Amazon(亚马逊公司)

Yanshen Sun, Jianfeng He, Limeng Cui, Shuo Lei, Chang-Tien Lu

AI总结:

针对LLM假新闻攻击依赖人工收集信息且易失真的问题,本文提出VLPrompt攻击方法并构建VLPFN数据集,通过多种检测器与人类研究评估其检测挑战。

摘要:

大型语言模型(LLM)近期的进展使假新闻的生成成为可能,尤其是在医疗保健等复杂领域。已有研究强调了在有人协助和无人协助情况下LLM生成假新闻的欺骗能力之间存在差距,但提示技术的潜力尚未得到充分探索。因此,本研究旨在确定提示策略能否有效缩小这一差距。当前基于LLM的假新闻攻击需要人工干预来收集信息,且常常遗漏细节并无法保持上下文一致性。为此,为更好地理解威胁策略,我们提出了一种强大的假新闻攻击方法,称为条件类变分自编码器提示(conditional Variational-autoencoder-Like Prompt,VLPrompt)。与现有方法不同,VLPrompt无需额外的数据收集,同时能够保持上下文连贯性并保留原文的复杂细节。为推动未来关于检测VLPrompt攻击的研究,我们创建了一个名为VLPrompt fake news(VLPFN)的新数据集,其中包含真实文本和虚假文本。我们开展了包括多种检测方法和新颖的人类研究指标在内的实验,以评估它们在该数据集上的性能,并得出了大量发现。

英文摘要:

Recent advancements in Large Language Models (LLMs) have enabled the creation of fake news, particularly in complex fields like healthcare. Studies highlight the gap in the deceptive power of LLM-generated fake news with and without human assistance, yet the potential of prompting techniques has not been fully explored. Thus, this work aims to determine whether prompting strategies can effectively narrow this gap. Current LLM-based fake news attacks require human intervention for information gathering and often miss details and fail to maintain context consistency. Therefore, to better understand threat tactics, we propose a strong fake news attack method called conditional Variational-autoencoder-Like Prompt (VLPrompt). Unlike current methods, VLPrompt eliminates the need for additional data collection while maintaining contextual coherence and preserving the intricacies of the original text. To propel future research on detecting VLPrompt attacks, we created a new dataset named VLPrompt fake news (VLPFN) containing real and fake texts. Our experiments, including various detection methods and novel human study metrics, were conducted to assess their performance on our dataset, yielding numerous findings.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑