Verification of ID of a person commiting an extremely cruel war crime during the Russian invasion of Ukraine.
Most importantly, identification of the person from an extremely graphic video disseminated via numerous Telegram channels and social media in Russia and Ukraine proved wrong the first ID made by bloggers. When the original video was posted in late July, Ukrainian bloggers immediately named a man from Kalmykia region (Russian Federation) to be the torturer and ferociously doxxed him. After The Insider and Bellingcat verification of the video, they proved to be wrong since the actual torturer was another person. Besides, it led to discussion on IDing people in the wartime and all in all usage of face recognition techniques, becoming a textbook example of the technology abuse.
Metrics: at least a million people learnt about the personality of this person, it was shared thousands of times (according to the views metrics and information about quotation in Ukrainian media and media in other countries).
Photo/video geolocation, face recognition, social networks profiling, ID verification
Context about the project:
The investigation revealed that the initial ID made by various media outlets and individuals online in the first couple days after the video was posted, was wrong. The doxxed person didn’t participate in the execution.
Reporting on the war crimes, investigators are exposed to vicarious trauma, and on the other hand they appear to be one of few people trained to ID places and people with a decent quality and speed.
What can other journalists learn from this project?
Double- and triple-check reports on sensitive issues. Not use face recognition technologies blindly and learn about their mechanism (which is known for being racist – which proved in this case). Train in geolocation and verification techniques.