The internet is being flooded with AI content. How can we tell what is human?

4 points

3 months ago

The internet is entering a strange phase. For two decades, the main challenge for digital platforms was managing an overwhelming amount of content produced by people. Today the nature of the problem has shifted. What is growing fastest is no longer the number of humans posting online, but the number of machines doing it.

Bots write articles, generate videos, compose music, post comments and even simulate debates. In many cases they do this faster, cheaper and at volumes no human team could realistically match. The result is a phenomenon that some researchers have begun to describe as “algorithmic pollution”: platforms saturated with synthetic material that looks authentic but does not come from human experience.

Inside technology companies and newsrooms, a question is quietly becoming unavoidable. How do you separate what humans made from what machines generated?

Behind the scenes of the internet, the race to answer that question has already started.

The invisible avalanche The numbers alone reveal how serious the situation has become. Music streaming platforms, for example, are receiving tens of thousands of tracks created with artificial intelligence every single day. In a recent disclosure, Deezer reported that it detects more than 60,000 AI-generated songs daily, representing roughly 39% of all uploads on the platform. According to the company, up to 85% of those tracks appear to be fraudulent, created to manipulate royalty systems or recommendation algorithms. https://www.theverge.com/news/870186/deezer-ai-music-detection-commercially-available

Music is only one part of the story. A report from the research group AI Forensics identified hundreds of automated accounts on TikTok producing synthetic content at industrial scale, generating billions of views every month. Many of these accounts post dozens of videos per day and rarely disclose that the material is artificial. https://www.theguardian.com/technology/2025/dec/03/anti-immigrant-material-among-ai-generated-content-getting-billions-of-views-on-tiktok

For platforms this creates a structural dilemma. Recommendation algorithms were designed to reward volume, engagement and consistency. Artificial intelligence excels at all three.

Left unchanged, the systems that decide what goes viral may end up promoting machines over people. Full content here: <https://chat-to.dev/post?id=K29WWDZVRmZmQ3IyNWVEaGo1WWl1dz09&redirect=/>

4 comments