TikTok Expands Tools to Help People Spot, Shape and Understand AI-Generated Content

TikTok today announced a focused set of updates designed to increase transparency around AI-generated content (AIGC), give people clearer choices about how much AIGC appears in their For You feed, and raise public understanding of responsible AI use. As AI-generated content continues to surge globally, TikTok aims to remain a leader in responsible adoption by ensuring that creative advances are matched with clear labeling, robust safeguards, and meaningful user control.

Accordingly, in the coming weeks TikTok will begin testing an AIGC control inside its existing ‘Manage Topics’ settings so people can choose how much AI-generated content they see. ‘Manage Topics’ already lets people tune the frequency of content by interest areas; the AIGC control works the same way. It serves as a personalization tool rather than a content-removal feature, enabling users to tailor their experience. Those who value AI-driven creative or educational material can surface more of it, while others may adjust their settings to see less. This choice supports discovery and user agency as the goal is to help people decide what kinds of content best match their expectations and values.

TikTok is also strengthening how it labels AI-generated media. The platform already requires creators to disclose realistic AI content and combines creator disclosures, internal detection models, and industry metadata standards to do so. One such industry standard is C2PA Content Credentials, which embeds provenance metadata into a file so platforms and other tools that support C2PA can read whether, when and how content was created or edited. These systems have helped TikTok apply AIGC labels to over 1.3 billion videos to date.

To add an extra layer of reliability, TikTok will begin applying an “invisible watermark” to AI-generated content created with TikTok’s tools and to content uploaded with C2PA credentials. This invisible watermark is a platform-readable signal embedded imperceptibly in the media; it is designed so that TikTok, using its own verification systems, can detect and read the watermark even when visible labels or metadata are stripped during reuploads or re-editing. Due to the watermark being embedded in ways that are not apparent to viewers, it is far harder for bad actors to remove or tamper with than a visible tag. The watermark therefore helps TikTok maintain consistent labeling and provides context about edits or transformations a piece of content has undergone, enabling more accurate enforcement and user information across reuploads.

These measures build on TikTok’s broader investments in trustworthy AI innovation, from creative tools that help people express themselves to AI systems that support safety and wellbeing across the platform. TikTok will continue to share updates as these programmes roll out and as the company works with industry partners to advance responsible practices for synthetic media.