How a Silent War on TikTok Uses AI to Spread Racist Content
Amsterdam, donderdag, 4 december 2025.
TikTok is being overwhelmed by billions of views driven by automated bots that distribute AI-generated videos — including false images of young girls and racist fake interviews with migrants. According to a French study, more than half of these videos are not marked as fake, despite the platform having AI labeling tools. Most concerning? The majority of these bots were created only at the beginning of 2025, indicating a rapidly growing and barely regulated phenomenon. This development shows just how easily AI can now be misused to spread misinformation — without users even noticing. The question remains: how much longer will this go unnoticed?
An Invisible Wave of AI Content Overwhelms TikTok
Since November 2025, automated bots have flooded TikTok with AI-generated videos and images that have attracted billions of views, according to research by the French NGO AI Forensics [1]. During a one-month investigation, nearly 44,000 AI-generated videos were posted by 350 bot accounts, amassing over 4.5 billion views in total [1]. Most of these accounts — two-thirds — were created in the first months of 2025, indicating a recent and rapid increase in this phenomenon [1]. These bots use AI to generate images of young girls appearing extremely young, underage, or dangerously thin, which many view as the sexualisation of minors [1]. The bots also spread fake videos of migrants appearing in fabricated interviews that promote racist stereotypes and glorify violence, including against police, with images of migrants ‘picking jobs’ [1]. The danger lies not only in the content but also in the fact that TikTok barely flags these videos as fake: more than half of the videos studied carried no AI label, despite the platform having the tools to do so [1].
The Failure of AI Labeling and the Role of Accountability
TikTok has 1.3 billion posts with an AI label on the platform, but the company states that only 1.0% of videos are labeled by TikTok itself with the warning ‘contains AI-generated media’ [1]. The remaining labels are often added by creators themselves, undermining their reliability [1]. AI Forensics warned of the risk that AI-generated images could be used to fuel anti-migration and racist sentiments, and recommended creating a dedicated AI section on the platform to clearly distinguish content created by real people from that produced by artificial intelligence [1]. The company has dismissed the research by AI Forensics and The Guardian as ‘unfounded’ [1]. The shift from human moderators to AI was introduced following the dismissal of hundreds of employees in Berlin, including Dutch and German moderators, weakening oversight of harmful content [1]. Plans to replace German staff with AI remain incomplete, but the threat persists [1].
From Manipulation to Political Tension: The Threat to Democracy
The spread of AI-generated content extends beyond racist imagery. In France, President Emmanuel Macron is launching a campaign to combat online disinformation, focusing on ethically labeled news broadcasts ahead of the 2027 presidential elections [2]. This campaign, active since November 2025, has been joined by regional newspapers and public gatherings [2]. However, right-wing media, led by billionaire Vincent Bolloré, has responded with strong resistance, including statements from CNews host Pascal Praud, who accused Macron of having an ‘authoritarian streak’ [2]. Journal du Dimanche claimed Macron aimed to ‘control media that do not think as he does’ [2]. These political tensions reveal how sensitive society is to the idea of state oversight of information, even when it is intended to fight fake news [2]. The U.S. State Department had planned to visit Paris on 2 December 2025, where it would emphasize the defence of ‘freedom of speech’ and ‘digital freedom,’ highlighting the international context of this debate [2].
The Human Response: Recognition and Awareness
Users are starting to notice when AI content is being used. On 3 December 2025, Instagram user adesope_shopsydoo posted: ‘This is FAKE NEWS …… THIS IS DEFINITELY AI …, would never EVER EVER in my wildest dreams write , tweet or say such about Wiz for that matter ., NEVER….. THIS IS AI 😱🤦🏾♂️’ [3]. His post, linked to a video of rapper Wiz, reflects a growing awareness among public figures [3]. On TikTok itself, users like Stacyalyse (username) are actively identifying AI content, although the platform does not explicitly designate her profile as a real person [4]. These individuals play a vital role in strengthening media literacy, though their efforts are insufficient without technological and platform-level support [1][3].