TikTok wants creators to label their AI videos

TikTok is rolling out a new toolset that lets creators label content generated or significantly modified by artificial intelligence.

This update follows the platform's recent introduction of AI-powered video filters and an image generator for creating unique backgrounds. To make things clearer, TikTok is renaming its in-house effects to include "AI" directly in the title.

Mandatory disclosures and automatic detection

The new labels aren't restricted to built-in filters. Creators can apply them to any upload created entirely by AI or heavily altered with third-party tools.

Under TikTok's updated policy, labeling isn't optional if the content includes realistic AI-generated images, audio, or video. The goal is to provide viewers with proper context and curb the spread of misleading media, though creators can also use text overlays, stickers, or captions to meet disclosure rules.

Since relying on creator honesty is inherently imperfect, TikTok confirmed it is developing an internal algorithm to automatically detect and flag AI-generated content. A release date for the automated system hasn't been set yet.

The initiative was developed alongside TikTok's safety advisory councils and academic experts. In addition, the platform committed to the Partnership on AI's Responsible Practices for Synthetic Media and is teaming up with the nonprofit Digital Moment to host community roundtables with young users.

Why it matters

For creators, the disclosure bar is officially moving up. If you use AI tools to generate lifelike visuals or voice clones, you now have to bake labeling into your posting workflow to avoid policy violations. Fortunately, adding a sticker or turning on a toggle is practically frictionless today.

For the broader social media landscape, this sets an unavoidable precedent. As generative media gets indistinguishable from reality, platforms that ignore provenance risk polluting their feeds with high-velocity misinformation. TikTok moving early puts pressure on rival platforms to formalize their own labeling frameworks.

In our view, making filter names transparent and giving users an official toggle is a smart first step, but manual self-reporting has obvious limits. Malicious actors pushing deepfakes simply won't opt in. TikTok's transparency push will ultimately succeed or fail based on how accurate and scalable its upcoming automatic detection software turns out to be.