TikTok Floods Its Feed with 3 Times More AI-Generated Content Than YouTube

A recent report by Kapwing, a video creation tools company, has raised alarms about the saturation of AI-generated content on social platforms. According to the study, 59% of videos TikTok shows to a new account in its 'For You' feed are considered 'AI slop' — a derogatory term for low-quality content produced by AI — a figure three times higher than the 21% detected on YouTube Shorts. This finding not only reveals a disparity between platforms but also raises deep questions about algorithmic moderation, the attention economy, and user trust in an increasingly automated digital ecosystem.

Methodology and Key Findings

Kapwing manually reviewed over 10,000 TikTok videos across 20 categories and conducted a separate test with new accounts, counting AI-generated content in the first 500 videos of the 'For You' feed. On YouTube, they replicated the same test with Shorts. The results are striking: out of 500 videos on TikTok, 294 were 'AI slop'; on YouTube, only 104 out of 500. The methodology, while not without biases (Kapwing sells AI video creation tools), offers a valuable snapshot by relying on direct human review and comparing platforms under similar conditions.

The most impacted category was Children: 57% of the 2,000 videos analyzed under that label were AI-generated. The tag with the highest concentration was #cartoonkids, where 97 out of every 100 videos were artificial. It was followed by #cartoons and #babysong at 83%, and #forkids at 79%. Other categories with high incidence were Science and Education (35%), Health (33%), and History (33%), where voiceover narration and visual illustrations are common. At the opposite end, categories requiring physical presence or practical demonstration showed minimal rates: Fashion (1.3%), Music (1.5%), and Fitness (1.6%).

This pattern suggests that 'AI slop' thrives where automation is easiest: synthetic voices, simple animations, and generic scripts. The children's category is particularly vulnerable because young children cannot distinguish between real and AI-generated content, and parents trust platforms to provide safe material. Additionally, the high production volume of these videos (often channels posting dozens per day) makes them cheap to generate and difficult for moderation systems to detect.

Implications for Users and Platforms

The study comes at a time when TikTok had already labeled 1.3 billion videos as AI-generated as of November 2024, according to the report. This suggests the platform is aware of the phenomenon, but its algorithms still prioritize this type of content, possibly due to its low production cost and high volume. For TikTok, 'AI slop' represents a double-edged sword: on one hand, it keeps users engaged with a constant stream of content; on the other, it degrades the quality of the experience and may drive away authentic creators.

For users, the immediate consequence is a lower-quality consumption experience, with repetitive videos, generic narratives, and, in the case of children, potentially inappropriate or misleading content. The Kids category, which should be the most protected, turns out to be the most contaminated. This has serious implications: previous studies have shown that AI-generated children's content can include disturbing elements or subliminal messages, and its proliferation makes it difficult for parents to find genuine educational material.

From a market perspective, the Kapwing report also reveals a competitive gap. YouTube, with its history of stricter policies against spam and misinformation, appears to have more effective control over 'AI slop' in Shorts. However, the difference may be due to YouTube having a more established creator ecosystem and an algorithm that penalizes low-quality content, while TikTok rewards volume and virality. This could lead to an exodus of advertisers concerned about the quality of the advertising environment.

Reactions and Context

The report has been published by Slashdot and picked up by Search Engine Journal. Although Kapwing has a commercial interest in highlighting the proliferation of AI-generated content (its tool allows creating it), the manual methodology and direct comparison between platforms lend some credibility to the data. However, it is important to note that the study only covers new accounts, which could skew results if TikTok's algorithm shows more generic content to users without a history. Additionally, the definition of 'AI slop' is subjective and may include content that uses AI legitimately (e.g., audio enhancement) without being deceptive.

This phenomenon is not new: in 2023, NewsGuard already warned about hundreds of YouTube channels generating children's videos with AI, many of which contained factual errors or inappropriate themes. However, the current scale — tripling YouTube — positions TikTok as the main vector of 'AI slop' in the digital ecosystem. The difference lies in the speed of adoption: TikTok has integrated generative AI tools (such as video effects and avatars) more aggressively, lowering the barrier to entry for malicious creators.

Compared to previous events, such as the misinformation crisis on Facebook in 2016 or the plague of fake videos on YouTube Kids in 2017, 'AI slop' represents a qualitatively different challenge. It is not just about fake content, but mass-generated content that mimics popular formats without adding value. Platforms have responded with 'AI-generated' labels (TikTok already labels 1.3 billion videos), but this does not solve the underlying problem: algorithms continue to amplify the cheapest content to produce.

What Should Users Do?

• Verify the authenticity of videos, especially in sensitive categories like health or education. Look for signs such as robotic movements, emotionless voices, or visual errors.
• Use labeling and reporting tools when identifying misleading content. TikTok allows reporting videos as 'AI-generated' if they are not labeled.
• Be aware that algorithms prioritize volume over quality, and adjust feed preferences if possible. On TikTok, you can use the 'Not interested' option to reduce similar content.
• For parents, actively supervise children's content and use parental controls that restrict categories or keywords.

In summary, the Kapwing study confirms that TikTok is the new epicenter of low-quality AI-generated content, with direct implications for user experience and child safety. Platforms will need to rethink their moderation and labeling policies if they want to maintain audience trust. Meanwhile, the responsibility falls on users to critically navigate an ecosystem where authenticity is increasingly scarce.