About

Norman Sadeh is a Professor in the School of Computer Science at Carnegie Mellon University (CMU). He has co-founded and co-directed several innovative graduate programs at CMU, including the Privacy Engineering Program, the PhD Program in Societal Computing, and the MBA track in Technology Strategy and Product Management. His current research interests encompass cybersecurity, online privacy, Human-AI Interaction, AI governance, mobile computing, the Internet of Things, user-oriented machine learning, language technologies, and semantic web technologies. Dr. Sadeh is recognized for his pioneering work on AI-based privacy enhancing technologies, such as privacy assistants, automated privacy compliance tools, and NLP-based privacy enhancing technologies. He has conducted foundational research on modeling people's privacy expectations and preferences, as well as privacy and security nudging. His work has influenced privacy-enhancing solutions at major companies including Apple, Google, and Facebook/Meta, and has informed privacy policy and regulatory activities at agencies like the Federal Trade Commission and the California Office of the Attorney General. He is also the lead designer of CMU's Privacy Infrastructure for the Internet of Things (IoT). Norman Sadeh is a successful entrepreneur, having been the founding CEO and chairman of Wombat Security Technologies, a company that defined the user-oriented cybersecurity market and was acquired by Proofpoint in 2018. Technologies developed by him and his colleagues are used globally to protect tens of millions of users against cybersecurity attacks, including employees at over 75% of Fortune 100 companies. Earlier in his career, Prof. Sadeh conducted seminal work in AI planning and scheduling, agent-based supply chain management, workflow management, and automated trading. His research introduced probabilistic models for constraint satisfaction problems and demonstrated the importance of modeling decentralized and competitive supply chains. Products based on this research were commercialized by organizations such as IBM, CACI, Raytheon, Mitsubishi, Boeing, and the US Army. His 2001 best-selling book on M-Commerce anticipated the emergence of smartphones and highlighted usability, security, and privacy challenges in mobile commerce. His work on the livehoods project, which used social media data to interpret dynamic city patterns, received a test of time award from the AAAI Conference on Web and Social Media. Additionally, his research on recognizing mobile user activities while minimizing battery life has influenced technologies in modern smartphones. In the late 1990s, he served as Chief Scientist of the European Union's e-Commerce initiative, a major program in cybersecurity and privacy research and policy. Norman Sadeh holds a Ph.D. in Computer Science from CMU, with a major in Artificial Intelligence and a minor in Operations Research, an MS in computer science from the University of Southern California, and a BS/MS in electrical engineering and applied physics from the Free University of Brussels. His research and views on cybersecurity, privacy, mobile, and IoT technologies have been widely covered in the press. Between 2008 and 2019, he was also a visiting professor at Hong Kong University.

Research topics

• Computer Science
• World Wide Web
• Internet privacy
• Computer Security
• Sociology
• Psychology
• Data science
• Human–computer interaction
• Advertising
• Regional science
• Business
• Geography
• Visual arts

Selected publications

• No Privacy without AI

  Communications of the ACM · 2026-04-13

  article1st authorCorresponding

  Why privacy at scale requires AI.

  PublisherDOI

• Privacy Settings of Third-Party Libraries in Android Apps: A Study of Facebook SDKs

  Proceedings on Privacy Enhancing Technologies · 2025-03-07 · 1 citations

  articleOpen accessSenior author

  Previous studies have demonstrated that privacy issues in mobile apps often stem from the integration of third-party libraries (TPLs). To shed light on factors that contribute to these issues, we investigate the privacy-related configuration choices available to and made by Android app developers who incorporate the Facebook Android SDK and Facebook Audience Network SDK in their apps. We compile these Facebook SDKs' privacy-related settings and their defaults. Employing a multi-method approach that integrates static and dynamic analysis, we analyze more than 6,000 popular apps to determine whether the apps incorporate Facebook SDKs and, if so, whether and how developers modify settings. Finally, we assess how these settings align with the privacy practices that developers disclose in the apps’ privacy labels and policies. We observe widespread inconsistencies between practices and disclosures in popular apps. These inconsistencies often stem from privacy settings, including a substantial number of cases in which apps retain default settings over alternatives that offer greater privacy. We observe fewer possible compliance issues in potentially child-directed apps, but issues persist even in these apps. We discuss remediation strategies that SDK and TPL providers could employ to help developers, particularly developers with fewer resources who rely heavily on SDKs. Our recommendations include aligning default privacy settings with data minimization principles and other conservative practices and making privacy-related SDK information both easier to find and harder to miss.

  PublisherOA PDFDOI

• Missing Pieces: How Do Designs that Expose Uncertainty Longitudinally Impact Trust in AI Decision Aids? An In Situ Study of Gig Drivers

  2025-06-23 · 3 citations

  articleOpen access
  PublisherDOI

• Can a Cybersecurity Question Answering Assistant Help Change User Behavior? An In Situ Study

  2025-01-01 · 1 citations

  articleOpen accessSenior author

  Human actions or lack thereof contribute to a large majority of cybersecurity incidents.Traditionally, when looking for advice on cybersecurity questions, people have turned to search engines or social sites like Reddit.The rapid adoption of chatbot technologies is offering a potentially more direct way of getting similar advice.Initial research suggests, however, that while chatbot answers to common cybersecurity questions tend to be fairly accurate, they may not be very effective as they often fall short on other desired qualities such as understandability, actionability, or motivational power.Research in this area thus far has been limited to the evaluation by researchers themselves on a small number of synthetic questions.This article reports on what we believe to be the first in situ evaluation of a cybersecurity Question Answering (QA) assistant.We also evaluate a prompt engineered to help the cybersecurity QA assistant generate more effective answers.The study involved a 10-day deployment of a cybersecurity QA assistant in the form of a Chrome extension.Collectively, participants (N=51) evaluated answers generated by the assistant to over 1,000 cybersecurity questions they submitted as part of their regular day-to-day activities.The results suggest that a majority of participants found the assistant useful and often took actions based on the answers they received.In particular, the study indicates that prompting successfully improved the effectiveness of answers and, in particular, the likelihood that users follow their recommendations (fraction of participants who actually followed the advice was 0.514 with prompting vs. 0.402 without prompting, p=4.61E-04), an impact on people's actual behavior.We provide a detailed analysis of data collected in this study, discuss their implications, and outline next steps in the development and deployment of effective cybersecurity QA assistants that offer the promise of changing actual user behavior and of reducing human-related security incidents.

  PublisherDOI

• Making Teams and Influencing Agents: Efficiently Coordinating Decision Trees for Interpretable Multi-Agent Reinforcement Learning

  Proceedings of the AAAI/ACM Conference on AI Ethics and Society · 2025-10-15

  articleOpen access

  Poor interpretability hinders the practical applicability of multi-agent reinforcement learning (MARL) policies. Deploying interpretable surrogates of uninterpretable policies enhances the safety and verifiability of MARL for real-world applications. However, if these surrogates are to interact directly with the environment within human supervisory frameworks, they must be both performant and computationally efficient. Prior work on interpretable MARL has either sacrificed performance for computational efficiency or computational efficiency for performance. To address this issue, we propose HYDRAVIPER, a decision tree-based interpretable MARL algorithm. HYDRAVIPER coordinates training between agents based on expected team performance, and adaptively allocates budgets for environment interaction to improve computational efficiency. Experiments on standard benchmark environments for multi-agent coordination and traffic signal control show that HYDRAVIPER matches the performance of state-of-the-art methods using a fraction of the runtime, and that it maintains a Pareto frontier of performance for different interaction budgets.

  PublisherOA PDFDOI

• Applying large language models to sanitize self-disclosure in user-generated content

  Applied Soft Computing · 2025-06-01

  articleSenior author
  PublisherDOI

• Out of the Past: An AI-Enabled Pipeline for Traffic Simulation from Noisy, Multimodal Detector Data and Stakeholder Feedback

  ArXiv.org · 2025-05-27

  preprintOpen access

  How can a traffic simulation be designed to faithfully reflect real-world traffic conditions? One crucial step is modeling the volume of traffic demand. But past demand modeling approaches have relied on unrealistic or suboptimal heuristics, and they have failed to adequately account for the effects of noisy and multimodal data on simulation outcomes. In this work, we integrate advances in AI to construct a three-step, end-to-end pipeline for systematically modeling traffic demand from detector data: computer vision for vehicle counting from noisy camera footage, combinatorial optimization for vehicle route generation from multimodal data, and large language models for iterative simulation refinement from natural language feedback. Using a road network from Strongsville, Ohio as a testbed, we show that our pipeline accurately captures the city's traffic patterns in a granular simulation. Beyond Strongsville, incorporating noise and multimodality makes our framework generalizable to municipalities with different levels of data and infrastructure availability.

  PublisherOA PDFDOI

• Data Retention Disclosures in the Google Play Store: Opacity Remains the Norm

  2024-07-08 · 3 citations

  articleSenior author

  Privacy policies serve as the primary channel through which users are informed about the handling of their personal data, as required by regulations such as the General Data Protection Regulation (GDPR). This paper presents an evaluation of Android applications' privacy policies, focusing on how they articulate and disclose data retention periods. In this paper, we introduce a systematic approach that leverages Large Language Models to evaluate GDPR compliance regarding data retention disclosure across a diverse sample of 2,235 apps, demonstrating the applicability of the method at scale. Our approach reports a 0.904 F1 score, validated with a ground truth dataset manually annotated by legal experts and publicly released. Results show that over half of the examined policies are potentially non-compliant, with a significant subset indicating indefinite data retention and a high ratio of overlapping retention periods on the same privacy policy. This lack of compliance implies that those policies either fail to specify a retention period or provide unclear criteria for determining how long user data is kept. Thus, our study highlights the critical need to improve the clarity and enforcement of privacy policy practices, laying the groundwork for more transparent data governance.

  PublisherDOI

• Sharing is Not Always Caring: Delving Into Personal Data Transfer Compliance in Android Apps

  IEEE Access · 2024-01-01 · 9 citations

  articleOpen accessSenior author

  In an era marked by ubiquitous reliance on mobile applications for nearly every need, the opacity of apps’ behavior poses significant threats to their users’ privacy. Although major data protection regulations require apps to disclose their data practices transparently, previous studies have pointed out difficulties in doing so. To further delve into this issue, this article describes an automated method to capture data-sharing practices in Android apps and assess their proper disclosure according to the EU General Data Protection Regulation. We applied the method to 9,000 random Android apps, unveiling an uncomfortable reality: over 80% of Android applications that transfer personal data off device potentially fail to meet GDPR transparency requirements. We further investigate the role of third-party libraries, shedding light on the source of this problem and pointing towards measures to address it.

  PublisherOA PDFDOI

• Missing Pieces: How Do Designs that Expose Uncertainty Longitudinally Impact Trust in AI Decision Aids? An In Situ Study of Gig Drivers

  arXiv (Cornell University) · 2024-04-09

  preprintOpen accessSenior author

  Decision aids based on artificial intelligence (AI) induce a wide range of outcomes when they are deployed in uncertain environments. In this paper, we investigate how users' trust in recommendations from an AI decision aid is impacted over time by designs that expose uncertainty in predicted outcomes. Unlike previous work, we focus on gig driving - a real-world, repeated decision-making context. We report on a longitudinal mixed-methods study ($n=51$) where we measured gig drivers' trust as they interacted with an AI-based schedule recommendation tool. Our results show that participants' trust in the tool was shaped by both their first impressions of its accuracy and their longitudinal interactions with it; and that task-aligned framings of uncertainty improved trust by allowing participants to incorporate uncertainty into their decision-making processes. Additionally, we observed that trust depended on their characteristics as drivers, underscoring the need for more in situ studies of AI decision aids.

  PublisherOA PDFDOI

Recent grants

• CT-T User-Controllable Security and Privacy for Pervasive Computing

  NSF · $1.1M · 2006–2011

• TC: Medium: Collaborative Research: User-Controllable Policy Learning

  NSF · $724k · 2009–2013

• SaTC: CORE: Medium: Collaborative: Contextual Integrity: From Theory to Practice

  NSF · $431k · 2018–2023

• TWC SBE: Option: Frontier: Collaborative: Towards Effective Web Privacy Notice and Choice: A Multi-Disciplinary Prospective

  NSF · $3.4M · 2013–2019

• SBE: Medium: Towards Personalized Privacy Assistants

  NSF · $499k · 2015–2020

Frequent coauthors

• Lorrie Faith Cranor

  Carnegie Mellon University

  50 shared

• Florian Schaub

  University of Michigan–Ann Arbor

  33 shared

• Michael Benisch

  30 shared

• Jason Hong

  Carnegie Mellon University

  30 shared

• Shomir Wilson

  26 shared

• Patrick Gage Kelley

  26 shared

• Alessandro Acquisti

  22 shared

• Fabien Gandon

  21 shared

Labs

• Mobile Commerce LabPI

  Researching new technologies and applying user-centered design principles in the development of solutions to reconcile context-awareness and privacy in mobile and pervasive computing environments.

Education

• Ph.D., Computer Science

  Carnegie Mellon University

  1990

• M.S., Computer Science

  Carnegie Mellon University

  1986

• B.S., Computer Science

  University of Pennsylvania

  1983