A major platform quietly changes its terms: Your AI content can be freely used to train models, but compensation is only $100

【CryptoWorld】Recently, a major tech platform updated its service agreement with quite a big move. They significantly expanded the scope of “user content,” directly including AI-generated prompts and outputs. Then, they issued themselves a global license—an royalty-free license—that allows them to do anything with your content, including feeding it into AI models for training, without paying any compensation.

Even more outrageous are the new clauses: anti-“jailbreaking,” anti-prompt injection, and a penalty of $15,000 for batch scraping of 1 million pieces of content within 24 hours. Want to sue? You have to go to court in Texas. The compensation cap is $100. Want to organize a class action? No, the clause explicitly prohibits it.

This move has angered many researchers and legal experts. The reason is straightforward: these restrictive clauses could not only block academic research but also prevent users from seeking legal remedies. Your creative content is used to train models for profit, and if something goes wrong, your chances of compensation are minimized. Essentially, this is a case of power imbalance.