Clearview AI, Ownership, and the Digital Face-Image in the Age of Facial Recognition Technology

Abstract

The American facial recognition company Clearview AI created an app that allows clients—including police departments and government agencies such as the FBI and Department of Defense—to use photos to search a database of over 30 billion photos that have been scraped from the Web. Clearview’s activities have raised alarm bells and fueled conversations regarding data ethics, privacy, and the ownership of personal data. Clearview AI deals specifically with a type of personal data called the face-image, or a two-dimensional, photographic representation of the real human face, as in a reflection or digital photo. The scholarship thus far has mostly neglected to explore the ramifications of the face specifically as a form of data and potential object of possession. To understand the face as data and the face as a potentially owned object, it will be necessary to develop how the face operates in terms of the ethical/social relations between humans and what happens when the face is captured in a static face-image and its data extracted through facial recognition technology. Many have speculated that the face acts as a window into the soul or as a canvas for the emotional and moral inner lives of human beings, but how does the face in all of its transparency and opacity operate when two distinct human consciousnesses encounter each other? Using Levinas and Sartre, I will argue that the face acts as a medium for discursive, ethical relations between humans that then becomes objectified upon being seen or captured (Morris 2015, 67). Furthermore, I will also argue that current facial recognition technology further transforms the objectified face-image by extracting its features and mining it for quantitative data. Given the status of the face, I recommend clear, specific policy delimiting when the processing of digital data is allowed and mandating consumer education about data processing, such that netizens enter digital spaces with complete information regarding the potential for the distribution of their data, including their face-image.