LLM-PBE: Assessing Data Privacy in Large Language Models
University of California, Berkeley(加州大学伯克利分校);University of Texas at Austin(德克萨斯大学奥斯汀分校);University of Illinois Urbana-Champaign(伊利诺伊大学厄巴纳-香槟分校);National University of Singapore(新加坡国立大学);Center for AI Safety(AI安全中心);University of Chicago(芝加哥大学)
Qinbin Li, Junyuan Hong, Chulin Xie, Jeffrey Tan, Rachel Xin, Junyi Hou, Xavier Yin, Zhun Wang, Dan Hendrycks, Zhangyang Wang, Bo Li, Bingsheng He, Dawn Song
Large Language Models (LLMs) have become integral to numerous domains, significantly advancing applications in data management, mining, and analysis. Their profound capabilities in processing and interpreting complex language data, however, bring to light pressing concerns regarding data privacy, especially the risk of unintentional training data leakage. Despite the critical nature of this issue, there has been no existing literature to offer a comprehensive assessment of data privacy risks in LLMs. Addressing this gap, our paper introduces LLM-PBE, a toolkit crafted specifically for the systematic evaluation of data privacy risks in LLMs. LLM-PBE is designed to analyze privacy across the entire lifecycle of LLMs, incorporating diverse attack and defense strategies, and handling various data types and metrics. Through detailed experimentation with multiple LLMs, LLM-PBE facilitates an in-depth exploration of data privacy concerns, shedding light on influential factors such as model size, data characteristics, and evolving temporal dimensions. This study not only enriches the understanding of privacy issues in LLMs but also serves as a vital resource for future research in the field. Aimed at enhancing the breadth of knowledge in this area, the findings, resources, and our full technical report are made available at https://llm-pbe.github.io/, providing an open platform for academic and practical advancements in LLM privacy assessment.
PII-Bench: Evaluating Query-Aware Privacy Protection Systems
Institute of Fintech, Fudan University(复旦大学金融科技学院);Shanghai Key Laboratory of Data Science, School of Computer Science, Fudan University(复旦大学数据科学上海重点实验室);Laboratory of Data Analytics and Security, Fudan University(复旦大学数据安全分析实验室)
Padding Ain't Enough: Assessing the Privacy Guarantees of Encrypted DNS
DNS over TLS(DoT)和DNS over HTTPS(DoH)加密DNS,通过向被动攻击者隐藏DNS解析来保护用户隐私。然而,过去的攻击表明,加密的DNS仍然对流量分析敏感。因此,RFC 8467建议在加密之前填充消息,这大大降低了加密流量的特性。在本文中,我们表明,填充单独是不足以对抗DNS流量分析。我们提出了一种新的流量分析方法,结合大小和时间信息来推断用户访问的网站纯粹基于加密和填充的DNS跟踪。为此,我们对DNS序列进行建模,以捕捉通常触发数十个DNS解析而不仅仅是单个DNS事务的网站的复杂性。一项基于Alexa前10 k网站的封闭世界评估显示,攻击者可以在80.2%的网站中对至少一半的测试跟踪进行去匿名化,甚至可以正确标记32.0%的网站的所有跟踪。我们的研究结果破坏了DoT/DoH中最先进的消息填充策略的隐私目标。我们的结论表明,成功的缓解这种攻击必须删除查询响应之间的到达时间间隔的熵。
Bridging Data Barriers among Participants: Assessing the Potential of Geoenergy through Federated Learning
Ningbo Institute of Digital Twin, Eastern Institute of Technology(宁波数字孪生研究院,东方理工大学);The Hong Kong Polytechnic University(香港理工大学);Shanghai Jiao Tong University(上海交通大学);Research Institute for Smart Energy, The Hong Kong Polytechnic University(香港理工大学智慧能源研究院)
Assessing Privacy Leakage in Synthetic 3-D PET Imaging using Transversal GAN
Data Science Institute, University of British Columbia(不列颠哥伦比亚大学数据科学研究所);BC Cancer Research Institute(不列颠哥伦比亚癌症研究所);University of British Columbia(不列颠哥伦比亚大学)
Assessing the Privacy Benefits of Domain Name Encryption
随着互联网用户越来越了解他们的互联网通信被观察的可能性,网络流量加密技术的使用(例如,HTTPS/TLS)正在崛起。但是,即使启用了加密,用户也会通过DNS查询和TLS的服务器名称指示(SNI)扩展泄漏有关其访问的域的信息。最近有两个改善这个问题的提案是DNS over HTTPS/TLS(DoH/DoT)和加密SNI(ESNI)。在本文中,我们的目标是评估这些建议的隐私利益,考虑主机名和IP地址之间的关系,后者仍然暴露。我们从全球九个有利位置进行DNS查询,以描述这种关系。我们量化了ESNI为不同的托管和CDN提供商提供的隐私增益,使用两个不同的指标,由于共同托管和IP地址变化的动态的k-匿名程度。我们发现,20%的域研究将不会获得任何隐私的好处,因为他们有一个一对一的映射之间的主机名和IP地址。另一方面,30%的域名将在k值大于100的情况下获得显著的隐私优势,因为这些域名与100多个其他域名共同托管。访问者隐私将得到有意义改善的域名远不受欢迎,而对于受欢迎的域名,好处并不显著。分析长期域名的IP地址动态,我们发现只有7.7%的域名每天都会更改其托管IP地址。最后,我们讨论了网站所有者和托管/CDN提供商的潜在方法,以最大限度地提高ESNI的隐私优势。
Tutela: An Open-Source Tool for Assessing User-Privacy on Ethereum and Tornado Cash
Stanford University(斯坦福大学);Eötvös Loránd University(罗兰大学);Convex Research;Universidad de Buenos Aires(布宜诺斯艾利斯大学);Facultad de Ingenieria(工程学院);LambdaClass
T2ISafety: Benchmark for Assessing Fairness, Toxicity, and Privacy in Image Generation
Shanghai Artificial Intelligence Laboratory(上海人工智能实验室);Beihang University(北航);Harbin Institute of Technology(哈尔滨工业大学);Chinese University of Hong Kong (Shenzhen)(香港中文大学(深圳));The University of Hong Kong(香港大学)
Unlocking Apple's Private Cloud Compute: An Analysis of Privacy-Preserving Artificial Intelligence
Hasso Plattner Institute, University of Potsdam(哈索普兰特纳研究所,波茨坦大学);TU Darmstadt, Secure Mobile Networking Lab(德累斯顿技术大学,安全移动网络实验室);IMDEA Networks Institute, Madrid, Spain(IMDEA网络研究所,马德里,西班牙)