📄 中文摘要
2023年，Reddit以6000万美元的价格将其整个历史内容档案出售给谷歌，而创作这些内容的用户却没有得到任何补偿，也没有被询问过。这一做法早在数年前就已开始。如今，所有主要的语言模型都在没有有效同意的情况下使用了这些数据。如何在法律、技术和伦理上解决这个问题，成为了AI时代的一个重要未解难题。
你从未同意训练将取代你的 AI
出处: You Never Consented to Training the AI That Will Replace You
发布: 2026年3月7日
📄 English Summary
You Never Consented to Training the AI That Will Replace You
In 2023, Reddit sold its entire historical content archive to Google for $60 million, leaving the millions of users who created that content without compensation or consent. This practice began years earlier. Today, every major language model is trained on data collected without meaningful consent. Addressing the legal, technical, and ethical implications of this issue has become one of the defining unresolved challenges of the AI era.