Skip to main content
eScholarship
Open Access Publications from the University of California

Scholarly Works (6 results)

  • Article
  • Peer Reviewed
UC Berkeley Previously Published Works (2019)
Intelligent voice assistants (IVAs) and other voice-enabled devices already form an integral component of the Internet of Things and will continue to grow in popularity. As their capabilities evolve, they will move beyond relying on the wake-words today's IVAs use, engaging instead in continuous listening. Though potentially useful, the continuous recording and analysis of speech can pose a serious threat to individuals' privacy. Ideally, users would be able to limit or control the types of information such devices have access to. But existing technical approaches are insufficient for enforcing any such restrictions. To begin formulating a solution, we develop a systematic methodology for studying continuous-listening applications and survey architectural approaches to designing a system that enhances privacy while preserving the benefits of always-listening assistants.
    Cover page: Privacy controls for always-listening devices
    • Article
    • Peer Reviewed
    UC Berkeley Previously Published Works (2018)
    Millions of smartphones are stolen in the United States every year, putting victims' personal information at risk since many users often do not lock their phones. To protect individuals' smartphones and the private data stored on them, we developed a system that automatically detects pickpocket and grab-and-run theft, in which a thief grabs the phone from a victim's hand then runs away. Our system applies machine learning to smartphone accelerometer data in order to detect possible theft incidents. Based on a field study and simulated theft scenarios, we are able to detect all thefts at a cost of 1 false alarm per week. Given that many smartphone users refuse to enable screen locking mechanisms over complaints that it takes too long to unlock their devices, our system could be used in conjunction with these systems in order to drastically decrease the number of times a user is asked to provide a lock code. That is, our system could be used to prompt smartphone users for PINs or passcodes only when theft events have been detected.
      Cover page: Detecting Phone Theft Using Machine Learning
      • Thesis
      • Peer Reviewed
      UC Berkeley Electronic Theses and Dissertations (2021)

      Intelligent voice assistants and other microphone-enabled Internet of Things devices are increasingly popular but may pose grave privacy risks. As more technologies take advantage of continuous audio access, new products may emerge that are passively listening and actively analyzing all conversations they hear, instead of only processing commands that follow their wake-words. The research in this dissertation investigates how to develop a privacy permission system for these devices. It takes a user-centered perspective on this problem and uses empirical methods to study the research questions that can guide the design of the privacy controls:

      - What are people's expectations about what information needs to be protected and in what context?- Which privacy-enhancing techniques could be feasibly applied to limit the devices' listening? How do people perceive their trade-offs and acceptability? - Which interfaces and affordances would allow users to express their privacy preferences and explore the implications of their choices?

      This dissertation explores these questions through a combination of surveys, user studies, and prototype evaluations. Major conclusions include:

      - People exhibit nuanced and heterogeneous preferences, notably in relation to other members of their households, and are especially wary of undisclosed data flows to third parties. They are most protective of financial data and other information that can cause them harm.- Block-listing and filtering approaches may be most feasible to implement. In combination with existing techniques and privacy-friendly design choices, they can address immediate user requirements. However, more complex privacy needs must be addressed with content-based controls, which require additional research in privacy and natural language understanding. - People appreciate the control install-time and runtime permissions provide over their own data. However, both have challenges with their user experience. Transparency-based approaches may be comparatively frictionless.

        Cover page: Privacy Controls for Always-Listening Devices
        • Article
        • Peer Reviewed
        UC Berkeley Previously Published Works (2017)
        Current smartphone operating systems regulate application permissions by prompting users on an ask-on-first-use basis. Prior research has shown that this method is ineffective because it fails to account for context: the circumstances under which an application first requests access to data may be vastly different than the circumstances under which it subsequently requests access. We performed a longitudinal 131-person field study to analyze the contextuality behind user privacy decisions to regulate access to sensitive resources. We built a classifier to make privacy decisions on the user's behalf by detecting when context has changed and, when necessary, inferring privacy preferences based on the user's past decisions and behavior. Our goal is to automatically grant appropriate resource requests without further user intervention, deny inappropriate requests, and only prompt the user when the system is uncertain of the user's preferences. We show that our approach can accurately predict users' privacy decisions 96.8% of the time, which is a four-fold reduction in error rate compared to current systems.
          • Article
          • Peer Reviewed
          UC Berkeley Previously Published Works (2012)
          Application platforms provide applications with accessto hardware (e.g., GPS and cameras) and personal data.Modern platforms use permission systems to protect access to these resources. The nature of these permissionsystems vary widely across platforms. Some platformsobtain user consent as part of installation, while othersdisplay runtime consent dialogs. We propose a set ofguidelines to aid platform designers in determining themost appropriate permission-granting mechanism for agiven permission. We apply our proposal to a smartphone platform. A preliminary evaluation indicates thatour model will reduce the number of warnings presentedto users, thereby reducing habituation effects.
            Cover page: How To Ask For Permission
            • Article
            • Peer Reviewed
            UC Berkeley Previously Published Works (2018)
            Modern mobile operating systems implement an ask-on-firstuse policy to regulate applications' access to private user data: the user is prompted to allow or deny access to a sensitive resource the first time an app attempts to use it. Prior research shows that this model may not adequately capture user privacy preferences because subsequent requests may occur under varying contexts. To address this shortcoming, we implemented a novel privacy management system in Android, in which we use contextual signals to build a classifier that predicts user privacy preferences under various scenarios. We performed a 37-person field study to evaluate this new permission model under normal device usage. From our exit interviews and collection of over 5 million data points from participants, we show that this new permission model reduces the error rate by 75% (i.e., fewer privacy violations), while preserving usability. We offer guidelines for how platforms can better support user privacy decision making.
              Cover page: Contextualizing Privacy Decisions for Better Prediction (and Protection)
              Top