TikTok is taking aim at deepfakes featuring public figures, launching a detection tool worldwide as the platform prepares for a series of European elections next year and seeks to blunt the spread of AI-generated disinformation.
The “likeness detection tool” is being introduced gradually. It will allow politicians and TikTok accounts with large followings to identify and report AI-generated material that uses their image without permission.
Deepfakes include text, images, video and audio that look or sound authentic but have been created or altered with artificial intelligence.
For TikTok, which is owned by China-based ByteDance, the tool represents a major part of its election-related defenses ahead of votes across Europe next year, including in Italy, France and Greece.
She said the likeness tool would help “politicians and government identify potentially AI-generated content that has been shared potentially without their consent.”
The European Union began introducing broad AI transparency rules in August, including a requirement that deepfakes and other AI-generated material carry labels.
For several years, TikTok has required creators to disclose realistic AI-generated images, audio and video.
Last year, the company said advances in technology would enable it to reduce the number of content-moderation jobs and locations around the world.
Artificial intelligence has become increasingly important in social-media moderation as platforms seek to reduce the cost of human reviews covering pornography, misinformation and hate speech.
YouTube has likewise built a likeness-detection tool that allows creators to find AI-generated videos portraying people impersonating them.