TikTok feeds now feel stacked with AI-generated clips: face-swapped celebrities, synthetic news explainers, AI-animated filters, fully generated “stories.” Until now, users mostly had to accept that blend. TikTok has started to change that. The platform is rolling out a control inside its Manage Topics settings that lets people increase or decrease how much AI-generated content appears in the For You feed, while it strengthens labeling and watermarking for synthetic media.
This move sits at the intersection of three pressures: user fatigue with AI “slop”, regulators pushing for deepfake transparency, and creators experimenting aggressively with generative tools. For security, trust-and-safety and policy teams, TikTok’s shift offers an early template for how large platforms may handle AI content at scale – and what it means when a platform gives users explicit TikTok AI content control instead of silently reshaping feeds.
𝗪𝗵𝗮𝘁 𝗧𝗶𝗸𝗧𝗼𝗸’𝘀 𝗻𝗲𝘄 𝗔𝗜 𝗰𝗼𝗻𝘁𝗿𝗼𝗹 𝗮𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝗱𝗼𝗲𝘀
TikTok is adding an AI-generated content (AIGC) slider to the same Manage Topics section that already lets people tune categories like Dance, Sports or Current Affairs. The control focuses on one thing: how often AI-generated videos appear in the For You feed.
Instead of burying AI behind hidden flags, TikTok presents AI as a topic in its own right. Users can:
-
turn the dial down to see fewer AI-generated videos in recommendations, or
-
turn it up if they actively enjoy stylised AI clips, filters and synthetic edits.
The control does not remove AI entirely. TikTok still reserves the right to show some AI-generated content, especially when it considers a video highly relevant or newsworthy. However, it finally acknowledges that AI generation is no longer just an effect – it is a content class that people should manage, just like any other topic.
𝗛𝗼𝘄 𝗧𝗶𝗸𝗧𝗼𝗸 𝗶𝗱𝗲𝗻𝘁𝗶𝗳𝗶𝗲𝘀 𝗔𝗜 𝗰𝗼𝗻𝘁𝗲𝗻𝘁 𝗶𝗻 𝘁𝗵𝗲 𝗳𝗶𝗿𝘀𝘁 𝗽𝗹𝗮𝗰𝗲
The slider only works if TikTok can reliably recognise synthetic media. To support that, the platform leans on a layered system:
First, TikTok requires creators to label realistic AI-generated content when it shows believable people, voices or scenes that never actually happened. Creators should mark videos that are either fully generated or significantly edited by AI – for example, deepfake speeches, swapped faces or heavily re-imagined footage.
Second, TikTok applies its own automatic labeling. When creators use TikTok’s AI effects, or when they upload content that carries Content Credentials from the C2PA industry standard, the platform can automatically tag the video as AI-generated. That auto-label appears in the interface so viewers know the clip involved AI.
Third, TikTok is rolling out “invisible” watermarking for AI content produced with its tools. Instead of a visible logo, the watermark lives in video metadata or subtle signal patterns. That approach aims to:
-
survive basic re-edits, crops or re-uploads, and
-
help TikTok and partner tools identify AI-generated material even after it travels off-platform.
Because of these layers, the new TikTok AI content control slider can work with billions of labeled videos rather than relying entirely on best-effort detection. TikTok already reports more than a billion AI-labeled clips, which shows how quickly synthetic media has become mainstream inside its ecosystem.
𝗧𝗶𝗴𝗵𝘁𝗲𝗿 𝗿𝘂𝗹𝗲𝘀 𝗳𝗼𝗿 𝗱𝗲𝗲𝗽𝗳𝗮𝗸𝗲𝘀 𝗮𝗻𝗱 𝗵𝗮𝗿𝗺𝗳𝘂𝗹 𝗔𝗜 𝗺𝗲𝗱𝗶𝗮
TikTok isn’t only adding knobs for users; it is also tightening its policy lines around AI content. The platform already:
-
bans AI content that uses the likeness of minors or private individuals without consent,
-
prohibits synthetic media that misleads people in harmful ways or impersonates others, and
-
enforces its broader rules on hate speech, harassment and misinformation regardless of whether a video uses AI.
Now, as it upgrades AI detection and control, TikTok pairs those rules with better enforcement tools. Stronger watermarking, C2PA metadata support and auto-labeling all feed into moderation and appeals systems. When a video crosses the line into harmful deepfake territory, TikTok can more easily show how it identified the clip as synthetic.
In parallel, TikTok has started to link AI policy to extremism controls, signalling that AI-generated propaganda, doctored extremist footage and synthetic recruitment material will face stricter scrutiny. That shift matters for security and OSINT practitioners who regularly rely on TikTok clips to track real-world events and threat narratives.
𝗪𝗵𝗮𝘁 𝗰𝗵𝗮𝗻𝗴𝗲𝘀 𝗳𝗼𝗿 𝗰𝗿𝗲𝗮𝘁𝗼𝗿𝘀 𝗮𝗻𝗱 𝗯𝗿𝗮𝗻𝗱𝘀 𝘂𝘀𝗶𝗻𝗴 𝗔𝗜
Creators who embraced AI as a shortcut now face more nuanced incentives. On one side, TikTok still rewards creative use of AI tools, especially when videos clearly label synthetic elements and avoid deceptive framing. On the other side, the platform hands viewers a dial that can deprioritise AI content at the recommendation level.
Because of that, brands and influencers need to think harder about how they disclose AI and where they use it. Over-reliance on generic AI clips risks landing in a bucket that many users actively tune down. Conversely, a small amount of well-signposted AI – for visualisation, accessibility or humour can still perform strongly when it respects context.
Mislabeling carries its own risk. If a creator slaps an AI label on normal footage as a gimmick, they may violate TikTok’s terms and draw moderator attention for the wrong reasons. Likewise, hiding the AI origin of a realistic political, news or celebrity clip increases the chance of removal and reputational damage.
𝗨𝘀𝗲𝗿 𝗰𝗵𝗼𝗶𝗰𝗲, 𝗺𝗶𝘀𝗶𝗻𝗳𝗼𝗿𝗺𝗮𝘁𝗶𝗼𝗻 𝗮𝗻𝗱 𝗮𝗹𝗴𝗼𝗿𝗶𝘁𝗵𝗺𝗶𝗰 𝘁𝗿𝘂𝘀𝘁
By exposing a TikTok AI content control knob directly to end users, the company sends a clear message: the platform recognises AI fatigue and wants people to feel some agency in the feed again. That move may also help TikTok defend itself in upcoming regulatory debates, since it can point to user-visible controls and robust labeling as evidence of “responsible AI.”
However, the control does not solve misinformation by itself. Even with labels and sliders, some users will still engage heavily with synthetic political memes or conspiracy-heavy AI animations. Research on content labels shows that warnings can increase awareness that something is AI-generated without fully changing engagement behaviour, especially in entertainment and polarised contexts.
Because of that, security and policy teams should treat TikTok’s new tools as partial mitigations. They help honest users avoid AI overload and spot synthetic media faster; they do not stop motivated actors from pushing deceptive content or communities from amplifying it.
𝗜𝗺𝗽𝗹𝗶𝗰𝗮𝘁𝗶𝗼𝗻𝘀 𝗳𝗼𝗿 𝘀𝗮𝗳𝗲𝘁𝘆, 𝗰𝗼𝗺𝗽𝗹𝗶𝗮𝗻𝗰𝗲 𝗮𝗻𝗱 𝗱𝗶𝗴𝗶𝘁𝗮𝗹 𝗿𝗶𝘀𝗸 𝘁𝗲𝗮𝗺𝘀
For organisations, the move carries several practical implications:
First, internal comms, marketing and brand-safety teams need to assume that audiences can see AI labels on corporate content and adjust their messaging accordingly. When a brand uses AI, transparency becomes part of trust management, not just a legal checkbox.
Second, security and fraud teams should expect AI-generated scams and impersonation content to keep using TikTok as a channel, even under stricter rules. Watermarks and labels improve detection, yet they do not fully block re-uploads, edited clips or off-platform sharing. Teams should continue to monitor TikTok for executive impersonation, fake giveaways and synthetic support scams that jump into direct messages and other apps.
Third, compliance and public-policy groups can treat TikTok’s changes as a signal for where regulation is heading. Multiple jurisdictions now move toward mandatory deepfake labeling, stricter age protections and more transparent content provenance. Platforms are responding with a mix of user controls, watermarking and moderation commitments. That pattern will likely repeat across other short-form and livestream platforms.
𝗔𝗜 𝗹𝗮𝗯𝗲𝗹𝘀 𝗮𝗻𝗱 𝘀𝗹𝗶𝗱𝗲𝗿𝘀 𝗮𝗿𝗲 𝗯𝗲𝗰𝗼𝗺𝗶𝗻𝗴 𝗽𝗹𝗮𝘁𝗳𝗼𝗿𝗺 𝘀𝘁𝗮𝗻𝗱𝗮𝗿𝗱
TikTok’s decision to give users direct control over how much AI-generated content they see, while strengthening labeling and watermarking, signals a new phase for social platforms. Synthetic media has moved from novelty to baseline. Now platforms need to prove that they can govern AI content at scale without suffocating creativity or letting deepfakes and synthetic propaganda run wild.
For everyday users, the new controls offer a simple, intuitive answer to “why is my feed full of AI videos?” For professionals in security, policy and brand protection, they act as an early model of what AI governance will look like in consumer apps over the next few years: a blend of user choice, cryptographic provenance and stricter enforcement around the most risky forms of synthetic media.
𝗙𝗔𝗤𝗦
What exactly does TikTok’s new AI slider control?
The new slider in TikTok’s Manage Topics settings lets users increase or decrease how often AI-generated content appears in their For You feed. It doesn’t remove AI videos entirely, but it gives people a way to push synthetic media higher or lower in their recommendations without digging through obscure menus.
Does TikTok completely block harmful AI content?
TikTok prohibits AI-generated content that shows minors or private individuals without consent and bans synthetic media that misleads people in harmful ways or impersonates others. Even when creators label their content as AI-generated, TikTok can still remove it if it violates broader policies on extremism, harassment or misinformation.
How does TikTok detect AI-generated videos?
TikTok combines creator self-labeling, automatic labels for content created with its AI effects and recognition of C2PA Content Credentials embedded in uploads. It also plans to add invisible watermarks to videos created with its own AI tools, so it can recognise them even after re-editing or re-uploading.
Will choosing “less AI” make my feed completely human-only?
No. The slider reduces the presence of AI-generated content in the For You feed but does not guarantee a fully AI-free experience. TikTok still considers factors like relevance, popularity and safety when it recommends videos and may show some AI content even when users turn the AI dial down.
What should brands and creators do differently because of these changes?
Brands and creators should label realistic AI-generated content accurately, avoid using synthetic media in misleading ways and assume that some viewers will actively reduce AI in their feeds. They should treat transparency about AI use as part of their trust strategy and monitor how these controls affect reach and engagement over time.
