TikTok tests a control to dial down AI-generated content

On November 18, 2025, TikTok announced a control that lets viewers choose how much AI-generated content appears in their For You feed, along with plans for invisible watermarks to make AI labels harder to strip. In July 2026 it went further and said it would test detection aimed at whole accounts that post AI-generated spam about politics, finance, and health.

The date above is the November 2025 announcement. TikTok described the feed control as a test, and we have not found an announcement that it is available to everyone.

What changed

The feed control. TikTok said it would start testing an AI-generated content setting in Manage topics, the menu where people already adjust how much they see of categories such as Dance or Sports (TikTok). TikTok says the setting tailors the mix of the feed rather than removing content: someone who likes AI-made history videos can see more, and someone who doesn’t can see less (TechCrunch).

Invisible watermarks. TikTok said it would start adding a watermark only it can read to AI content made with its own tools, such as AI Editor Pro, and to uploads that carry C2PA Content Credentials. The aim is to keep labels attached after content is downloaded, edited, or re-uploaded (TikTok).

Labeling rules. TikTok’s AI-generated content rules already required creators to label AI content that shows realistic people or scenes. TikTok can also apply an “AI-generated” label automatically, for example when content carries Content Credentials. Unlabeled realistic AI content can be removed, restricted, or labeled by TikTok.

July 2026 update. On July 10, 2026, TikTok said it would test improved detection for “accounts dedicated to posting AI-generated spam” on three high-risk topics: politics and current events, financial advice, and medical content. The test works at the account level, not video by video (Search Engine Journal). That month, TikTok also said it had labeled more than 3 billion pieces of content as AI-generated. On July 27, 2026, the C2PA, the group behind Content Credentials, announced that TikTok had joined its steering committee and repeated the 3 billion figure (C2PA).

What it means for faceless channels

Impact is medium. None of this bans AI content or removes monetization by itself. But it changes who sees AI-made videos, and the July test targets the riskiest faceless niches directly.

• Reach becomes viewer-chosen. Viewers who have the control can turn labeled AI content down. A faceless channel whose appeal rests on “AI-made” novelty is the most exposed. A channel people watch for the story, the research, or the craft is the least exposed.
• Labels are harder to lose. Watermarks and Content Credentials mean AI content made with common tools is more likely to be labeled whether or not you label it. Label it yourself.
• Finance, health, and politics carry account-level risk. An account that posts only templated AI videos in these topics is what the July test describes. Action against a whole account is a different order of risk from a single removed video.

The craft answer is the same as on YouTube: be the channel viewers would choose even with the slider turned down. The Standard’s original angle, researched script, and disclosure and compliance criteria apply directly.

What to do now

• Label realistic AI content with TikTok’s AI-generated content setting when you post. Never strip watermarks or Content Credentials. See AI disclosure rules.
• If you post about money, health, or politics, cite sources on screen, avoid templated scripts, and make it clear who is behind the account. See the personal finance niche guide.
• Do not run an account on one template. Each video needs its own research and edit. Changing the surface format does not change what the July test describes.
• Make sure each video meets the Creator Rewards originality rules if you earn on TikTok.

What we don’t know yet

• How many viewers use the slider, and whether lowering AI content affects labeled videos only or also content TikTok detects without a label.
• What counts as an AI-spam account: TikTok has not published the criteria for the July 2026 test, or what happens to an account it flags.
• Availability: TikTok described the feed control as a test in November 2025. We have not found a TikTok announcement that it has reached all users or all markets.