TikTok Is Quietly Testing AI Likeness Detection for US Creators

1. A real-time live selfie scan — the creator takes a selfie in the moment, confirming liveness.
2. A government ID document upload — a driver's license, passport, or other official ID.

TikTok states it does not retain the submitted ID documents or facial biometric data . The verification is handled by Jumio and used only to confirm that the person enrolling is who they claim to be. On Jumio's side, a "reference selfie" is typically stored on its servers and associated with the user's account depending on the organization's data retention settings , but TikTok's policy is that it does not keep those images itself .

Once verified, TikTok's AI begins scanning the platform for any uploaded content that appears to use the creator's likeness. If potential matches are found, the creator can review them and flag unauthorized AI-generated impersonations directly to TikTok for takedown .

Current Availability: A Narrow US Pilot

The test is currently limited to a small subset of US creators only . TikTok has not announced a broader rollout timeline or any international expansion .

How TikTok Compares to YouTube's Likeness Detection

YouTube's system is significantly more mature and widely accessible:

YouTube began testing Likeness Detection in 2024 with YouTube Partner Program creators, then expanded it to public figures and the entertainment industry through early 2026 . On May 18, 2026, YouTube announced the tool was available to all users aged 18 and above with a YouTube account . YouTube describes it as a matching service similar to Content ID, giving creators the ability to find, review, and request removal of AI-generated content that uses their face . The key difference is scale and access: YouTube's tool is open to hundreds of millions of adult users, while TikTok's remains a narrow US-only pilot with no announced timeline for broader release.

Broader Context: The AI Flood on TikTok

TikTok has labeled over 1.3 billion AI-generated videos on its platform, with 52% of all TikTok content now involving some form of AI — from filters and effects to full AI avatars . An estimated 7.5 million AI-generated videos are uploaded daily, representing about 22% of all uploads . The platform has already introduced mandatory AI content labeling and a feed-control feature letting users reduce AI content visibility .

The EU AI Act Deadline Looms

This TikTok test arrives just weeks before the EU AI Act's Article 50 transparency obligations take effect on August 2, 2026 . That rule requires all AI-generated or manipulated content — deepfakes, synthetic audio, images, video, text — to be marked in a machine-readable format and visibly labeled when deployed in the EU . On May 8, 2026, the European Commission published draft guidelines specifying that even deepfake content created without intent to deceive must be labeled . The Code of Practice on Transparency of AI-Generated Content, facilitated by the Commission, further reinforces these marking requirements . Platforms operating in the EU — including TikTok and YouTube — will need robust detection, labeling, and enforcement systems to comply by that date.

TikTok's test can be seen as a proactive step toward those compliance requirements, but with only weeks until the EU deadline, it remains a limited pilot in one market rather than a finished enforcement system covering the platform's global user base.