A paper by CyLab researchers about how people perceive advanced video analytics has received the Future of Privacy Forum's annual Privacy Papers for Policymakers Award. The paper is one of six to receive the award and be identified as "must-read" privacy papers of the year for policy makers.

The paper, "Did You Know This Camera Tracks Your Mood? Understanding Privacy Expectations and Preferences in the Age of Video Analytics," was originally published and presented at the 2021 Privacy Enhancing Technologies Symposium this past summer by CyLab's Aerin Zhang, a Ph.D. student in CMU's School of Computer Science.

"Cameras are everywhere and increasingly coupled with video analytics software that can identify our faces, record attendance at events, track our moods and more," Zhang said. "The rapid deployment of video analytics across ever more diverse contexts calls for a better understanding of how people feel about these deployments, including their expectations to be notified about and to be able to exercise control over associated data practices."

In the paper, Zhang and her co-authors presented results from a study designed to capture people's privacy expectations and preferences when confronted with diverse, realistic uses of video analytics as part of their regular daily activities. They evaluated the extent to which people expect the use of video analytics in different contexts, how they feel about them, how they would want to be notified about them, and what privacy choices they might want to have — such as the ability to restrict the collection and use of their footage.

"People generally did not realize that video analytics could be used for such a diverse set of purposes at such a diverse set of venues and how powerful the technology can be," Zhang said. "Our study also shows that people's privacy expectations and preferences are diverse, too, and vary significantly across different deployment scenarios."

Yuanyuan Feng, a co-author of the paper who was a postdoctoral researcher in SCS at the time of its writing, said their study is the first to capture participants' responses to a wide range of realistic video analytics deployments in the context of their daily lives.

"Over the span of 10 days, we gathered over 2,000 detailed responses to different video analytics deployment scenarios from over 120 participants," Feng said.

CyLab's Norman Sadeh, the study's principal investigator and a professor in SCS, said the state-of-the-art today involves notifying data subjects by merely placing signs that read "this area is under camera surveillance."

"This clearly falls short," Sadeh said. "It falls short of disclosing a lot of information that matters to people and does not provide them with any practical form of control over the collection and use of their footage, as mandated by regulations such as GDPR or CCPA/CPRA under at least some contexts."

The authors found that many — though not all — people seem to have grown accustomed to the use of some video surveillance technologies. But many express surprise and a desire to be informed about and exercise some control over more recent ways video is used, such as measuring productivity in the workplace, in marketing, for attendance-tracking purposes, or when making inferences about someone's health, including mental health.

The paper ends with a discussion on the implications of these findings in terms of deploying more effective mechanisms for notice and choice around video analytics.

"We advocate for the development of interfaces that simplify the task of managing notices and configuring controls," Sadeh said. "The development of such interfaces would, however, require the adoption of standards for notification and for people to communicate their opt-in/opt-out choices to video analytics operators, something that in turn would likely require new regulation."

The Privacy Papers for Policymakers Award recognizes leading privacy scholarship that is relevant to policymakers in Congress, at federal agencies and for data protection authorities abroad.

Winning authors will present their work at an annual event with top policymakers and privacy leaders in February.

The research reported in this paper was conducted as part of the Personalized Privacy Assistant Project, including work on a privacy infrastructure for the internet of things. This work was funded under National Science Foundation grants and a DARPA Brandeis privacy research grant.