Deepfakes, Forensic Science, and Police Investigations
By Joseph Schafer
USA Today recently ran a piece on the emergence of “deepfakes” https://www.usatoday.com/videos/news/2019/04/26/deepfakes-detection-have-you-been-tricked-fake-obama/3588955002/
Deepfakes is a term applied to the ability to manipulate video to modify words and possible actions. In other words, to take something that is ostensibly real and modify it in such a way that the video conveys something entirely different. The implications for policing, while they might seem distant and rare, are profound, particularly when coupled with social media and a 280-character news cycle based on short attention spans and limited critical evaluation of sources.
The technology is being advanced, in part, by entertainment media. Video of an actor might be modified in post-production to correct an error or insert a better joke. An actor who has died can still complete their appearance in a film or TV show (although there might be legal, contractual, and financial implications).
Consider this technology in the hands of a foreign nation, however. Just days before an election, video might be released that seems to show a candidate making a particular statement. The capacity to interfere with free elections is profound and the risk in upcoming election cycles is astonishingly real. What was a pipe-dream in 2016 increasingly appears to be a reality for 2020.
In time, the risks here will not be limited to entertainment media or nations leveraging influence campaigns against each other. Imagine controversial police use of force event captured by a bystander’s mobile phone. In the near future, it might be possible to manipulate that video to make it appear the officer made biased, vulgar, or profane statements. In time, it might be possible to manipulate the video even more, to edit out citizen resistance or elevate the apparent force used by an officer.
In all of these examples, is anyone calling for the development of forensic expertise to analyze video and determine manipulation has taken place? Do crime labs and investigative agencies employ personnel with the requisite skill set for such analysis? How long will it take to develop credentialing standards for such forensic examiners? Will society care, or label reports that video has been altered “fake news”, continuing to believe that what they saw in a video with their own eyes represents reality?
Questions abound, but answers and solutions (for now) appear elusive. As future thinking police leaders, are we doing enough to call for attention and action on this issue before matters escalate beyond mitigation?