These videos were generated by AI within seconds or minutes, and the contributors have accounts on various social media platforms and post a large number of videos per day. Kebler likens the mass posting of AI-generated videos by these accounts to a 'brute force attack' on the Internet.

A brute force attack is a hacking technique that tries every possible combination of numbers and letters to crack a password, and although it is inefficient, it will eventually reach the correct answer. Similarly, mass posting of generative AI videos is a way to find 'videos recommended by SNS algorithms' by posting all kinds of videos, and when a video that suits the algorithm is found, it can be posted and used to earn advertising revenue.

What Kebler emphasizes here is that the intended 'viewers' of AI-generated videos are not humans, but rather social media algorithms. Because the posting span of AI-generated videos is overwhelmingly shorter than that of videos created by humans, content created by human creators is already being drowned out from the platform. In addition, because AI-generated videos are quicker to adapt to algorithms, Kebler points out that eventually the information ecosystem will collapse and the 'reality' people see online will also collapse.

Of course, creators active on YouTube, Instagram, TikTok, etc. have been trying to make videos that are likely to be recommended by the platform's algorithms for some time. However, no matter how much they analyze the algorithm to create videos, it is a relatively time-consuming process, and the number of videos that can be posted is about one every few days to weeks. In contrast, AI-generated video generation takes only a few seconds to a few minutes per video, so AI-generated video contributors can post hundreds to thousands of videos per day.

Kebler says he receives various marketing emails almost every day. YouTuber

Daniel Bitton , who has been trying to sell him products, claims that it's pointless to spend time creating high-quality YouTube videos, and that it's much more profitable to post a large number of AI-generated videos.

Bitton said he creates eight to 10 AI-generated videos in 30 minutes or less. In an email, he wrote, 'The great thing about posting short videos is that you can use AI, which essentially does 90% of the work. All you need to do is provide some criteria, press a button, create the video and let the algorithm do the work.' 'YouTube doesn't care about your production value. They are hungry to feed their audience, and their audiences are hungry for short content.'

Musa Mustafa, who, like Bitton, sells ways to make money on TikTok, YouTube and Instagram, said, 'AI is literally trained on the best content, so you get better results. If you want to create content and make money online, use AI. It's simple.'

Both Bitton and Mustafa argue that there's no point in spending big studios, huge amounts of money, long shoots and meticulous editing to create videos. Mustafa explains why by comparing social media users to hungry shoppers going to a convenience store at 2 a.m.: 'If you're hungry at 2 a.m., a crappy hot dog tastes better than a Michelin meal that you can finish in two bites. Well, TikTok works in the same way. Viewers don't expect or want a perfectly polished video.'

Mustafa added, 'When was the last time you saw a viral TikTok video and thought, 'Wow, that color is amazing!' You probably never, because no one cares. This is good news for you because it means you can focus on quantity instead of quality and make money.' He argues that crude AI-generated videos are enough to satisfy social media users and make money.

Furthermore, Kebler pointed out that while in normal brute force attacks, the attacked party strengthens its security, the situation is different in brute force attacks of AI-generated videos against platforms. Because the purpose of the platform is to attract users and have them watch advertisements, if users end up watching on the platform, there is no need to actively crack down on poorly generated AI videos. In other words, rather than being harmed by AI-generated videos, the platform is benefiting from them.

Meta CEO Mark Zuckerberg also said in a conference call in October 2024, 'I think we'll see entirely new categories of content, whether it's AI-generated or AI-summarized, or AI-curated content that already exists. I think that's going to be really exciting for Facebook and for Instagram and for Threads and other feed experiences.' It seems he has no intention of cracking down on AI-generated videos.

Meta also offers an AI tool for advertisers called ' Advantage+ ', which allows advertisers to generate many different versions of ads and conduct large-scale testing to select the most effective ads. Meta's AI tools for advertisers are a growing field, and at its earnings briefing in January , it was reported that more than 4 million advertisers are using at least one of its generative AI ad tools.

Many of the AI-generated videos that Kebler sees have been played millions of times and have many comments, many of which are of disgust or irritation. However, if you post such a comment or leave the video open while writing one, the algorithm learns that 'this video is attracting people's interest and should be recommended.' Kebler pointed out that 'this exploits a weakness in Instagram's algorithm, which sees engagement as a positive signal, and the people who create this type of content know it.'

If AI-generated videos continue to be left alone, we will see a future where AI-generated videos of golden retrievers are recommended to golden retriever owners, AI-generated videos of conspiracy theories are recommended to people who like apocalyptic theories and conspiracy theories, and AI-generated videos of Jesus Christ and the Pope are recommended to religious people. And in today's world, where social media is playing a greater role in our lives, videos recommended by algorithms affect our perception of the real world, so AI-generated videos will also affect reality, Kebler argued.