Meta removes terms language on scraping for AI training

Meta has removed language from its commercial terms that critics said could have given it broad permission to scrape websites, including news domains, for AI training. The Verge reported that the change followed backlash from publishers and other web content owners. The original wording referred to content from domains and web pages, which raised alarms among news organizations already worried about AI systems using their work without payment or clear permission. The concerns go beyond legal language. Publishers are watching traffic from search and social platforms become less reliable, while AI-generated summaries and chatbots increasingly answer questions without sending users to original sources. Any term that appears to authorize large-scale scraping can therefore feel like an existential threat. Meta's revision may reduce immediate confusion, but it does not settle the larger fight over scraping, licensing, fair use, and compensation for content used to train AI models. The episode also shows how public pressure can shape corporate policy. Even if Meta's terms are now narrower, the underlying tension remains: AI companies want data, while publishers want control and revenue. Governments and courts are still working through the rules. For now, the removal is a tactical retreat rather than a final resolution, and content owners are likely to keep scrutinizing AI training policies for any sign of overreach.