A Zelensky Deepfake Was Quickly Defeated. The Next One Might Not Be

 Article: “A Zelensky Deepfake Was Quickly Defeated. The Next One Might Not Be” - https://www.wired.com/story/zelensky-deepfake-facebook-twitter-playbook/

This month we saw the first weaponized use of deepfakes during an armed conflict. A deepfake was released of an imitation Zelensky (he was motionless and had an unusual sounding voice) telling Ukrainians to lay down their weapons.  This video emerged not only on Facebook and Youtube but also Telegram and Russian social network VKontakte.

TV Channel Ukraine 24 was hacked and the deepfake video and a summary of the fake news appeared on its website. Minutes after the fake news appeared on TV, Zelensky posted a Facebook video stating the video was fake.  Soon after, the head of security policy at Meta tweeted that they had removed the deepfake video for violating its policy against “misleading manipulated media.” A Twitter spokesperson provided a statement committing to tracking the video and removing it where it violated rules banning “deceptive synthetic media.” A YouTube spokesperson also communicated their removal of the deepfake video uploads. 

The unfolding of events shows that under the right conditions, deepfakes can be vanquished. However, it was, in a sense, “easy” for Zelensky to defeat the deepfake – his government had prepared for such a scenario, he is one of the highest-profile people in the world, and the deepfake was of poor quality.  Other political leaders in conflicts may be less fortunate and thus more vulnerable to deepfakes. 

In class, we have evaluated the responsibility of social media in regulating fake news and the role of advertising and machine learning in skewing reality.  While technology is underway to create automatic deepfake detectors, it is not fully developed yet.  How do we ensure the tools of digital marketing (from algorithms affecting search to social media) do not allow deepfakes to become political weapons?