More than a fifth of the videos that YouTube shows who opens an account first They are what several researchers already define as "AI garbage": automatically generated pieces of low quality and with hardly any narrative, designed almost exclusively to scrape together clicks, retention and advertising money.
The phenomenon, far from being marginal, points to a global automated content industry which sneaks into the platform's initial recommendations and reaches millions of users worldwide, with a notable impact also in European countries and in Spain, where tens of millions of people already follow this type of channel.
An experiment that reveals the weight of AI junk in recommendations
The video editing company Kapwing conducted a simple but very revealing test: Create a new YouTube account and see what the algorithm recommends. In that first contact with the platform, of the first 500 suggested videos, 104 were AI-generated content of very questionable quality, that is, just over 20% of the total.
Furthermore, the researchers identified that about a third of those 500 videos could be classified as “meaningless” or “destructive” content, a category that included both automated productions and other low-quality clips whose sole purpose is to retain attention at all costs.
This type of video is characterized by minimal or non-existent arguments, jarring images, and repetitive plotsdesigned to engage a wide range of users, including children, and to make the most of the platform's automatic recommendations.
The results of the experiment, reported by media outlets such as The GuardianThey fuel the debate about To what extent does the YouTube algorithm prioritize quality? versus the simple ability to generate views and playback time.
Experts consulted in the study warn that this “AI junk” not only floods the initial recommendations, but also It can influence how new users understand the platform, upon first encountering a catalog full of noise and empty messages.
15.000 channels analyzed and a new automated content industry
Beyond the experiment with the new account, Kapwing also analyzed The 15.000 most popular YouTube channels in the worldincluding the top 100 in each country. In that sample, they found 278 channels that exclusively publish videos generated with artificial intelligence, categorized as "AI garbage."
In total, these 278 channels accumulate over 63.000 billion views and 221 million subscribersThese figures give an idea of the magnitude of the phenomenon. Estimated revenues are around $117 million annually, demonstrating that the business model is, currently, highly profitable.
Kapwing describes this trend as the rise of a global industry of automated, addictive, and almost context-free contentwhich is not only present on YouTube, but also on TikTok, Facebook, X and other major social networks.
These productions often utilize animation templates, synthetic voices, and AI-generated scripts to manufacture, at high speed, and edit and convert videos with 4K quality, serialized videos that are repeated with slight variationsbut the algorithm may continue to promote them if it detects that they keep the audience engaged.
The fact that a significant portion of the fastest-growing channels on the platform are based on this type of content suggests, according to researchers, that the pressure to maximize visits and advertising revenue It outweighs incentives to promote quality informative or creative content.
Spain and other countries: millions of followers of AI channels
The same study details that the audience of these automated channels is widely distributed across the planet. In Spain, it is estimated that around 20 million users follow one of these AI channels., a considerable figure for a country with just over 47 million inhabitants.
Other markets also show a massive reach: 18 million followers in Egypt, 14,5 million in the United States, and 13,5 million in Brazilamong others. Although the report does not cover all European countries, it suggests that this type of content is very widespread in the region.
The combination of very low barriers to entry—all you need is a computer, an internet connection, and accessible AI tools—with the global audience potential offered by YouTube has multiplied the speed of expansion of these formats.
In the European case, where there are open debates about platform regulation and AI, the proliferation of low-quality automated videos It adds a new front: the protection of minors, digital mental health, and the quality of the information ecosystem.
For Spanish users, this means that when opening an account or consuming content casually, It's relatively easy to end up caught in a chain of recommendations dominated by artificial, repetitive productions with little real value beyond immediate entertainment.
Bandar Apna Dost and other success stories of AI garbage
Among the channels analyzed, the most viewed is Bandar Apna Dost, based in India, which has around 2.400 billion viewsHis AI-generated videos revolve around the adventures of a talking monkey and a muscular character clearly inspired by the Hulk, who face fantastic creatures or embark on outlandish situations such as traveling in a helicopter built from tomatoes.
Kapwing estimates that this channel could be generating annual income of up to $4,25 millionAll this thanks to a formula based on eye-catching animations, cartoonish violence and absurd plots, with hardly any narrative depth.
Technology and digital rights researcher Rohini Lakshané suggests that the popularity of these videos may be related to the exaggerated nature of its characters, the hypermasculinization of the main figures, and the lack of a complex narrativeThis makes the content understandable for viewers of any age or educational level.
Another example cited in the study is Pouty Frenchie, a Singapore-based channel which has reportedly amassed nearly 2.000 billion views. His videos depict a French bulldog living imaginary adventures, often accompanied by childlike laughter and absurd situations, also generated using AI tools.
In this case, the estimates speak of approximately $4 million in annual revenueBoth Bandar Apna Dost and Pouty Frenchie illustrate how the combination of easily recognizable characters, fast pace, and massive production can become a money-making machine on YouTube.
Entire communities dedicated to exploiting the algorithm
Journalist Max Read, who has been investigating these phenomena for some time, explains that Specific communities have emerged on Telegram, WhatsApp, Discord, and online forums where tips, guides and templates are shared for creating AI-related garbage videos that are eye-catching enough to go viral.
These spaces sell courses that teach how to identify profitable niches, optimize titles and thumbnails, or take advantage of the latest image and video generation tools to produce large volumes of content with very few resources.
Some of these niches can even seem surreal: Read mentions, for example, AI-generated videos of pressure cookers exploding in the kitchendesigned solely to attract user curiosity and accumulate visits without offering useful information or context.
According to the journalist, many of the creators who dedicate themselves to this come from countries with Relatively low average salaries compared to potential YouTube earningsand they see the mass production of automated content as a fast track to improving their economic situation.
This environment of “content factories” reinforces the impression that these are not isolated cases, but rather a parallel economy around the recommendation algorithmwhere the important thing is understanding how to hook him, not so much offering something valuable to the viewer.
Concern for user experience and information quality
The experts consulted by Kapwing and by the media outlets that have disseminated the study agree that the rise of AI junk is deteriorating the user experience on platforms like YouTubeFor a user arriving with no prior history, the first impression may be of an environment saturated with noisy, repetitive, and context-free videos.
Furthermore, a warning is issued regarding the risk of “information pollution”By mixing educational or informative content with misleading, absurd, or even potentially harmful pieces in the same flow of recommendations, the viewer is not always clear about what they are seeing or why it is being shown to them.
In Europe, where there is intense debate about the regulation of AI and the responsibility of large platforms, phenomena like this raise questions about How automated content is managed and what controls should be applied when they are directed at mass audiences, including minors.
Organizations dedicated to defending digital rights insist that this is not just a problem of “quality” in an aesthetic sense, but of a possible impact on attention, perception of reality, and trust in online information sources.
The general feeling among analysts is that, if the logic behind the recommendation is based almost exclusively on maximizing viewing time, The space for more elaborate and critical content is being displaced., with consequences for public debate and for the diversity of voices present on the platform.
YouTube's response: AI as a neutral tool
In response to the study's findings, a YouTube spokesperson emphasized that the Artificial intelligence is a tool that can be used to create both high-quality and low-quality content.and that, in itself, does not determine the value of the videos.
The company insists that its priority remains Connecting users with reliable, high-quality content and that removes videos that violate community guidelines, regardless of whether they were created by people, machines, or a combination of both.
However, the figures compiled by Kapwing and other analyses, such as the one published by The Guardian, point to Almost 10% of the fastest-growing channels on the platform use AI-generated contentThis suggests that current measures may not be slowing the advance of automated waste disposal.
In this context, the debate opens up about whether it would be necessary for YouTube to offer greater transparency regarding the origin of recommended contentallowing the user to know if a video has been partially generated by AI or if it is part of a mass production pattern.
It also raises the question of whether the platform should adjust its recommendation system to penalize the proliferation of empty or misleading contentespecially when targeting particularly vulnerable audiences or when suggestions are dominated by those who have not yet shaped their own interests.
With all these elements on the table, the picture painted by Kapwing's study is one of a YouTube where AI junk has gained significant traction in initial recommendations.Supported by a global industry that knows how to exploit the algorithm and by a gigantic audience spread across countries like Spain and the rest of Europe, this forces a rethinking of the balance between freedom of creation, advertising business and responsibility towards users.
Madrid