Assistance, Please! Smartphone Users Prefer Help with Privacy Settings

Sorting out privacy settings for apps on a smartphone is no job for a mere mortal. A research project at Carnegie Mellon University found that most people preferred the settings offered up by an automated "personalized privacy assistant" compared to the ones they could figure out for themselves.

Reported in the paper, "Follow My Recommendations: A Personalized Privacy Assistant for Mobile App Permissions," the field study worked with 84 people using Android devices. Nearly 4 in 5 recommendations (79 percent) made by the assistant were adopted by users. After the initial set-up the same users received daily "nudges," encouraging them to continue reviewing and modifying their settings; but that follow-up encouragement only resulted in changes to five percent of the originally recommended settings, suggesting that people found them "useful and usable."

The assistant worked by engaging in a "dialog" with the user to understand overall comfort level in granting permissions to apps in certain categories. Through those answers, the app identified a user profile that was the best fit for a given person. The app privacy configuration recommendations were based on that profile. One profile "cluster," for example, described a person who denied location and contact access of social and finance apps. Another profile denied phone call log permission more often.

The project found that compared to users who were part of a control group the majority of recommendations in the assisted group were accepted, and participants kept most of the settings through the course of the study.

As the researchers concluded in their paper, "Our results suggest that personalized privacy assistants can indeed help users better manage their mobile app permission settings."

While the experiment focused on automated assistance for mobile interactions, they added, "We expect that personalized privacy assistant approaches [could] also be applied to support privacy decision making in other domains where privacy configuration or awareness is an issue."

"It's clear that people just can't cope with the complexities of privacy settings associated with the apps they have on their smartphones," said Norman Sadeh, professor of computer science, in an article on the experiment. "And it's not just smartphone apps. The growing number of sensors and other smart devices that make up the so-called internet of things will impact privacy and make it even more challenging for users to retain control over their data and how it is being used."

Sadeh has long been involved in researching privacy settings for social sites. Last year, he led a project to understand the impact of "mobile app privacy nudging."

The results of the latest study were presented at the Symposium on Usable Privacy and Security (SOUPS) in Denver, where it won the "SOUPS Privacy Award." The paper begins on page 36 of the proceedings PDF file.

About the Author

Dian Schaffhauser is a former senior contributing editor for 1105 Media's education publications THE Journal, Campus Technology and Spaces4Learning.

Featured

  • white clouds in the sky overlaid with glowing network nodes, circuits, and AI symbols

    AWS, Microsoft, Google, Others Make DeepSeek-R1 AI Model Available on Their Platforms

    Leading cloud service providers are now making the open source DeepSeek-R1 reasoning model available on their platforms, including Amazon, Microsoft, and Google.

  • illustration with geometric shapes, digital circuitry, and subtle icons of an open book, graduation cap, and lightbulb

    University of Michigan Launches Agentic AI Virtual Teaching Assistant

    At the University of Michigan's Stephen M. Ross School of Business, a new Virtual Teaching Assistant pilot program is utilizing agentic AI to provide students with 24/7 access to support and self-directed learning.

  • robot waving

    Copilot Updates Aim to Make AI More Personal

    Microsoft has unveiled a range of updates to its Copilot platform, marking a new phase in its effort to deliver what it calls a "true AI companion" that adapts to individual users' needs, preferences and routines.

  • modern college building with circuit and brain motifs

    Anthropic Launches Claude for Education

    Anthropic has announced a version of its Claude AI assistant tailored for higher education institutions. Claude for Education "gives academic institutions secure, reliable AI access for their entire community," the company said, to enable colleges and universities to develop and implement AI-enabled approaches across teaching, learning, and administration.