TikTok is going after AI-created deepfakes by prohibiting their use on private individuals.
The social media platform made the change in a refresh to TikTok’s community guidelines, which will also require users to prominently disclose when a deepfake is used in a video.
The new policy promises to prevent AI-created videos from misleading the public and harassing targeted users. By tapping the latest AI algorithms, deepfakes can swap a person’s face over your own, or take existing footage and manipulate the subject's lips to say something else.
In 2020, TikTok took an initial step to stop deepfakes by prohibiting “synthetic or manipulated content that misleads users by distorting the truth of events in a way that could cause harm.” But the restriction was broad and vague. Tuesday’s refresh of the community guidelines now clearly bans deepfakes of minors and adult private figures. It also prohibits deepfakes of public figures for political or commercial purposes.
However, a TikTok user can still create a deepfake involving a public figure “in certain contexts,” such as for artistic and educational purposes. This could also include deepfaking a celebrity doing a popular dance. So TikTok channels such as the Chinese Elon Musk “Yi long ma,” should be able to stick around, provided they disclose the use of deepfakes.
The new community guidelines will take effect on April 21. TIkTok added: “Over the coming months, we will provide additional training to our moderators in order to help enforce these updated rules and standards effectively as they start to roll out.”
To enforce the guidelines, TikTok says it will remove any content found breaking the rules. It can also temporarily or permanently ban accounts in “severe or repeated on-platform violations.”