Professors Amiri Sani and Tsudik Investigate Techniques to Authenticate Photos and Videos
Computer Science Professors Ardalan Amiri Sani and Gene Tsudik recently started work on a new project, “Verifiable Provenance and Subject Awareness for Photos and Videos.” Funded through the National Security Agency (NSA), the goal is to investigate secure, user-friendly techniques for establishing the origin and authenticity of digital photographs and video clips, and for ensuring the subject was aware of the process and context of the recording.
As outlined in their project proposal, people are increasingly using photos and videos in security-sensitive applications, such as citizen journalism, legal platforms, e-voting and video conferencing. “Indeed, the 2020 COVID-19 outbreak and its resulting social distancing approach accelerated the use of some of these applications, especially videoconferencing as well as electronic signatures and contracts,” note Amiri Sani and Tsudik. “Meanwhile, the recent increase in social unrest raises the importance and popularity of everyday people taking part in documenting history via ‘citizen journalism.’”
Amiri Sani and Tsudik identify two related security concerns. First is the lack of verifiable content provenance information, leading to an inability to authenticate origin and data integrity. This shortcoming enables manipulated or deep-fake content to be used to mislead the viewers. Second is the lack of verifiable evidence that the subject was aware of the recording process. An example they provide is of an attacker secretly recording someone and then using the video clip to impersonate the victim in signing a legal contract.
“Current systems may try to protect the confidentiality of these photos and videos through encryption, but confidentiality is not the only requirement,” says Amiri Sani. “In fact, in some cases (such as a video shared publicly), it might not even be critical.” He goes on to highlight the need to answer the following three questions about a photo or video: “Was it captured by a real camera, what modifications and image filters were applied to it after capture, and was the subject in it aware of being recorded?”
Amiri Sani explains that “the ability to answer these questions helps the consumers of a photo or a video have confidence that it wasn’t spoofed, reused or modified in ways that would change its intended purpose.”
The researchers will focus on designing and building a comprehensive framework for verifiable provenance and subject awareness for photos and videos. The two-year, $300,000 NSA grant will support this work.
— Shani Murray