Viral Deepfake Demo Prompts ByteDance to Limit New AI Video Tool - Sixth Tone
ByteDance limited features in its new generative video model, Seedance 2.0, after a user showed it could recreate a person's likeness from a single photo, raising deepfake concerns.
- Incident date
- May 2026
- Target
- Stephen Chow
ByteDance's new AI video tool, Seedance 2.0, faced restrictions shortly after its release due to deepfake concerns. A demonstration showed the tool could recreate a person's likeness, voice, and environment from minimal input.
What happened
An online influencer demonstrated that Seedance 2.0 could recreate his voice, office, and body using just one photo. The resulting video quickly went viral, generating millions of views and sparking debate about the tool's potential impact. Users created short clips, including AI-generated fight scenes featuring Stephen Chow, which led to questions about copyright infringement. ByteDance responded by restricting the use of real-person reference materials, blocking uploads of celebrity faces, and requiring identity verification for self-generated content. Despite the restrictions, Seedance 2.0 is being rolled out gradually, with higher-tier memberships unlocking advanced features.