Douyin Rectifies Violations, Takes Down 538,000 AI-Infringing Short Videos
Chinese short video platform Douyin announced it has taken down 538,000 AI-infringing videos this year. It acknowledged ongoing challenges in identifying sophisticated AI-generated content and IP misuses.
📋 Article Processing Timeline
- 📰 Published: April 23, 2026 at 15:52
- 🔍 Collected: April 23, 2026 at 16:02 (9 min after Published)
- 🤖 AI Analyzed: April 23, 2026 at 17:02 (1h 0m after Collected)
Central News Agency
(CNA Reporter Li Ya-wen, Shanghai, 23rd) Chinese short video social platform Douyin stated today that it has taken down 538,000 AI-infringing short videos so far this year. Douyin admitted that while the platform continues to focus on the issue of copyright-infringing works, it still faces dilemmas in its governance capabilities, such as the difficulty in judging AI-generated content and insufficient recognition ability.
Douyin issued a special governance announcement today on "AI-Generated Inappropriate Content and Infringement of Others' Rights." The announcement pointed out that Douyin will continue to strengthen the governance of AI-infringing works in 2026. As of now, it has cumulatively taken down 538,000 AI-infringing short videos and penalized more than 4,000 violating accounts.
Douyin's governance focuses on "violations such as face-swapping and voice-stealing using AI technology" and "malicious modification of classics through AI technology to spread bad content."
To prevent infringement acts such as the AI misappropriation of celebrities' portraits, Douyin will continue to improve its sample database. Combining large model recognition, technical inspections, and reporting information, it will promptly handle AI-infringing content related to rights holders within the sample database. Severe actions will be taken against violating content such as bloody, violent, pornographic, and vulgar material transmitted through the "malicious modification" of traditional literary and artistic works.
Douyin admitted that current rectification work generally faces difficulties such as the hard-to-judge nature of AI-generated content and insufficient AI voice recognition capability. Some violating works conduct secondary creation by combining real footage with AI-generated content, making it difficult to distinguish true from false and exposing limitations in technical recognition. It is also difficult to obtain authorization information at the first moment for specific characters and IP images generated by AI, which brings difficulties to the platform's judgment and handling work.
The Paper reported on the 23rd that some people had previously reflected that many AI short dramas used AI synthesis technology to misappropriate the portrait and voice of artist Jackson Yee (Yi Yangqianxi) without authorization, using them for drama creation and commercial monetization. Jackson Yee's supporters strongly urged the platform to take down and delete the infringing content, and the related news attracted attention.
The Cyberspace Administration of China (CAC) recently requested platforms to implement the labeling of AI-generated works. The CAC stated in March that some short videos containing fictional interpretations, staged marketing, or AI generation were not labeled, misleading public perception. It demanded platforms to fully implement labeling and stated that platforms failing to do so would be severely punished. (Editor: Chou Hui-ying) 1150423
Choose to stand with facts. Every sponsorship from you is the power to protect press freedom.
Download the Central News Agency "First Hand News" APP to grasp the latest news instantly.
The text, images, and audio/video on this website may not be reproduced, publicly broadcast, publicly transmitted, or utilized without authorization.
(CNA Reporter Li Ya-wen, Shanghai, 23rd) Chinese short video social platform Douyin stated today that it has taken down 538,000 AI-infringing short videos so far this year. Douyin admitted that while the platform continues to focus on the issue of copyright-infringing works, it still faces dilemmas in its governance capabilities, such as the difficulty in judging AI-generated content and insufficient recognition ability.
Douyin issued a special governance announcement today on "AI-Generated Inappropriate Content and Infringement of Others' Rights." The announcement pointed out that Douyin will continue to strengthen the governance of AI-infringing works in 2026. As of now, it has cumulatively taken down 538,000 AI-infringing short videos and penalized more than 4,000 violating accounts.
Douyin's governance focuses on "violations such as face-swapping and voice-stealing using AI technology" and "malicious modification of classics through AI technology to spread bad content."
To prevent infringement acts such as the AI misappropriation of celebrities' portraits, Douyin will continue to improve its sample database. Combining large model recognition, technical inspections, and reporting information, it will promptly handle AI-infringing content related to rights holders within the sample database. Severe actions will be taken against violating content such as bloody, violent, pornographic, and vulgar material transmitted through the "malicious modification" of traditional literary and artistic works.
Douyin admitted that current rectification work generally faces difficulties such as the hard-to-judge nature of AI-generated content and insufficient AI voice recognition capability. Some violating works conduct secondary creation by combining real footage with AI-generated content, making it difficult to distinguish true from false and exposing limitations in technical recognition. It is also difficult to obtain authorization information at the first moment for specific characters and IP images generated by AI, which brings difficulties to the platform's judgment and handling work.
The Paper reported on the 23rd that some people had previously reflected that many AI short dramas used AI synthesis technology to misappropriate the portrait and voice of artist Jackson Yee (Yi Yangqianxi) without authorization, using them for drama creation and commercial monetization. Jackson Yee's supporters strongly urged the platform to take down and delete the infringing content, and the related news attracted attention.
The Cyberspace Administration of China (CAC) recently requested platforms to implement the labeling of AI-generated works. The CAC stated in March that some short videos containing fictional interpretations, staged marketing, or AI generation were not labeled, misleading public perception. It demanded platforms to fully implement labeling and stated that platforms failing to do so would be severely punished. (Editor: Chou Hui-ying) 1150423
Choose to stand with facts. Every sponsorship from you is the power to protect press freedom.
Download the Central News Agency "First Hand News" APP to grasp the latest news instantly.
The text, images, and audio/video on this website may not be reproduced, publicly broadcast, publicly transmitted, or utilized without authorization.