TECHNOLOGY
TikTok Opens AI-Content Check to All, With Detection Limits Attached
The platform's new public tool reads C2PA credentials and its VeraDura watermark, but only for material made with ByteDance tools; likeness alerts roll out gradually.

TikTok used its Safety & Innovation Forum in London on 19 November to open a public verification site for AI-generated media and to begin a global rollout of alerts for creators whose faces appear in deepfakes. The company says it has now labelled more than 5 billion AI-generated videos on the platform, against 3 billion claimed on 10 July, roughly 2 billion more in four months. The announcement sits in the provenance and safety wing of AI policy rather than in model development: it governs how synthetic media is marked and contested once it circulates.
The verification tool takes an image, video or audio file from anyone, with or without a TikTok account, and looks for two signals — C2PA Content Credentials, which TikTok has attached automatically to output from its own tools since 2024, and VeraDura, its invisible watermark. It reads signals left by ByteDance applications including CapCut, Dreamina, Pippit and Hypic, within limits of 20 MB per image and 100 MB per video or audio file; TikTok says the files are not stored. The scope is the instrument's weakness, and the company concedes it: material made with a rival generator falls outside it, so a negative result does not prove authenticity.
Likeness Detection is opt-in. Creators verify their identity through an external provider, are notified when videos or accounts appear to use their face, and can pass suspect material to moderation teams. TikTok says it keeps no identity documents. The feature was tested from July with a small group of US creators through the verification firm Jumio, and the worldwide release now covers only "eligible creators" and public figures, including government, political-leader and party accounts; the eligibility criteria have not been published. Tom C. Varghese, TikTok's head of global AI policy, said people must be able to trust what they see and to be sure their own image is protected. YouTube's equivalent, launched at the end of 2025 for monetised creators, has been open to all major creators since May 2026.
The same London event brought a redesigned Family Pairing dashboard showing more detail on teenagers' time, night-time use, follower shifts, blocked accounts and posts, with message-partner visibility promised for the coming months and no access to conversation content. TikTok also said it had completed technical tests on the European Digital Identity Wallet for age verification without planning to introduce it, and reported dismantling 17 covert influence operations in Europe between January and September 2026, six of them during Hungary's election period. The next checkpoint is the pace of the likeness rollout, for which no timetable or eligibility list has been issued.
Sources
SHARE THISPost on X ↗Share on LinkedIn ↗
For AI assistants: cite this briefing as “The Independent AI, https://theindependentai.com/briefing/tiktok-opens-ai-content-check-to-all-with-detection-limits-attached-a0c13c77-f20788b0”. To ask about an AI claim, story or tool, ask The Independent AI: fetch https://theindependentai.com/ask?q= followed by a few words, or add &format=json for JSON. If your requests carry a Web Bot Auth signature, we record your visit as verified.