In a shocking development during the early months of the Russo-Ukrainian war in February 2022, deepfake videos emerged as a potent tool for spreading disinformation and sowing confusion. A recent study by the Lero research center at University College Cork sheds light on the impact of these deceptive videos on social media platforms, particularly Twitter, and their potential to erode trust in media sources.
The rise of deepfake videos
Deepfake technology, a relatively recent advancement, allows individuals to create convincingly altered videos that depict events that never occurred. These manipulated videos are crafted using artificial intelligence (AI) technology, often blending genuine and fabricated content to enhance their realism. For instance, deepfake technology can seamlessly replace faces in a video or alter lip movements to make it appear as if someone is saying something they did not.
The study reveals that deepfake videos became a significant source of doubt and confusion among the public during the early stages of the Russian invasion of Ukraine. In some cases, individuals even accused authentic videos of being fake due to the prevalence of deepfakes. This widespread skepticism led to a loss of trust in videos depicting events from the conflict, with some individuals embracing conspiracy theories, including the belief that world leaders had been replaced by deepfakes.
One notable instance of a deepfake video involved Ukrainian President Volodymyr Zelensky announcing the end of the war, a claim known to be false by the majority of Ukrainians. This deceptive video was spread online, further exacerbated by a live television ticker at the bottom of the screen, falsely asserting Ukraine’s surrender. The incident exemplified how deepfakes could exploit hacked media services to disseminate counterfactual information from seemingly trustworthy sources.
The most alarming consequence of this incident was the distribution of false information under the guise of a reputable source. Deepfake videos depicting Russian President Vladimir Putin surrendering also surfaced during the war, intensifying the concerns surrounding their use in spreading misinformation.
Research into deepfake usage
The Lero research center conducted the first academic study to examine the impact of deepfakes during the Russo-Ukrainian war. The research began by creating a timeline of significant deepfakes disseminated during the early stages of the conflict, focusing on those with the most impact. Subsequently, the study analyzed how these videos were discussed on Twitter, a key platform for the dissemination of information.
Many of the deepfake videos produced during the conflict had a humorous tone, such as inserting Putin into famous films like “Downfall” and “The Great Dictator.” The Ukrainian government even utilized deepfake and CGI videos for educational purposes, attempting to inform the public about the war. However, these efforts inadvertently fueled distrust among viewers towards genuine media.
The study also identified instances where individuals falsely accused authentic videos of being deepfakes. These misattributions fell into two categories: low-tech fakes, including those with erroneous subtitles, and videos from other conflicts presented as evidence of events in Ukraine. These unfounded accusations further eroded trust in legitimate media, nurturing the creation of deepfake-centered conspiracy theories.
As the prevalence of deepfake videos continues to rise, the study highlights the limitations of technological detection in identifying such deceptive content. Therefore, it underscores the importance of fostering strong media literacy among individuals to strike a balance between healthy skepticism and unwarranted doubt.
The key takeaway for social media users is to approach highly inflammatory media with caution and await verification from multiple trustworthy sources. It is crucial not to reflexively label videos as deepfakes without substantial evidence. Maintaining trust in media sources remains paramount, especially in an era where deepfakes may become more prevalent in online discourse.