TikTok confirmed the removal of AI 18+ videos after Dan Tri's feedback

(Dan Tri) - TikTok confirmed that AI videos with sensitive content that Dan Tri reported violated policies and were removed from the platform, and expanded its review of similar content.

After Dan Tri reflected on the situation of AI 18+ movie advertisements appearing on social networks, TikTok Vietnam representative gave an official response on the censorship and handling of this content.

Accordingly, the platform reviewed the videos reported by reporters and expanded its inspection of similar content.

"The results confirmed that these videos violated the platform's policies and have been removed," a TikTok representative informed.

The AI ​​18+ videos reported by Dan Tri violated policies and were removed from the platform (Photo: Doan Thuy).

The social network asserts that its Community Standards and Advertising Policies clearly state what types of content are and are not allowed. However, this unit has not announced the specific number of AI 18+ videos removed after the review.

Nearly 18 million violating ads were removed in 3 months

The platform representative said that from January to March 2026, this social network removed nearly 18 million ads that violated policies globally.

This is the total number of violative ads in many different content groups, not the number of AI 18+ movie ads.

As for sensitive content, the platform rules out nudity, sexual activity, sexual services, or sexually suggestive behavior involving teenagers. Some limited cases may be considered depending on the educational, documentary, fictional or artistic context.

The Advertising Policy also prohibits sexually suggestive or sexually explicit content, including nudity, sexual conduct or content, and sexually suggestive language and sounds.

Nearly 18 million policy-violating ads were removed globally in the first 3 months of 2026 (Screenshot).

Notably, the platform said that ads belonging to the default short film category are only delivered to users 18 years of age or older.

To determine age, this unit uses many measures, including self-declared date of birth information and age prediction models to detect accounts that likely belong to underage people.

In addition to handling violative content, the social network also applies its own regulations to videos that are allowed to exist but are not suitable for widespread recommendation.

Accordingly, some content may be limited from appearing on the Recommendations message board. Restrictions are also applied to mature or complex topics to reduce accessibility for teenage users.

Users can use tools like keyword filtering, restricted mode, or muting their accounts to adjust what appears in their feed.

Short film videos created by AI appear on social networks, some of which has sexual images and details and leads viewers to external applications (Photo: Doan Thuy).

Previously, some users reported to Dan Tri that they continuously encountered AI movie ads with 18+ elements even though they did not actively search for this type of content.

Dan Tri also asked the platform to clarify whether the act of accidentally stopping watching a sensitive video could become a signal that causes the system to continue recommending similar content.

However, this response does not directly explain the above signal processing mechanism.

More than 3 billion AI assets have been labeled

Facing the development of generative AI, the platform said it is investing in technology to identify, label and censor content created by AI.

According to provided data, to date, more than 3 billion pieces of content on social networks have been labeled as AI-generated products through labeling technologies and tools.

This number reflects the total amount of AI-labeled content, but does not necessarily mean it violates policies.

For AI-generated features provided by the platform, content must go through multiple rounds of censorship to prevent commands, reference documents or products from being created that show signs of violation.

The unit also integrates feedback channels for users to report unusual or inaccurate experiences when using AI features.

According to the platform, more than 3 billion pieces of content have been labeled with AI through content identification and labeling technologies and tools (Photo: Doan Thuy).

As for teenage users, the platform said it will limit access to new generative AI features in the early stages of deployment. This measure is intended to create more time to monitor the impact before deciding to expand the scope of supply.

The social network also rolled out parental controls, from time limits to custom filters for certain types of content.

The confirmation of removal of the reported videos shows that this content has passed censorship before being detected and handled.

However, the platform has not provided information about when the videos started appearing, the number of views before being removed, or specific measures to prevent similar versions from continuing to be posted.

In the context of AI making it increasingly easier to create and edit videos, the ability to early detect infringing content, limit its spread and prevent access by minor users are still issues that need to be addressed by platforms.