Overview
- TikTok said it will begin testing an invisible watermark in the coming weeks to tag AI‑generated videos, readable only by TikTok to deter removal.
- A new Manage Topics slider will let users set how frequently AI‑generated content appears in their For You feed, with a gradual rollout over the next few weeks.
- TikTok reports more than 1.3 billion videos have been labeled as AI‑generated since 2023 using creator disclosures and C2PA Content Credentials, which can be lost when re‑uploaded.
- Company data for the first half of 2025 shows about 86% of violating content was removed automatically, with significant reductions in inaccurate moderation decisions and human review.
- The company announced a $2 million AI literacy fund and detailed the measures at its European safety forum in Dublin, citing a rise in AI‑enabled scams and large takedowns in TikTok Shop.