Meta Scrubs the Word "Train" From Its Commercial Terms Amid AI Backlash

Meta has revised the commercial terms that govern how it uses content for artificial intelligence, and the most visible change is a missing word. The company removed "train" from the language, a edit that follows a stretch of public criticism over how Meta gathers material for its AI systems.

The wording change matters because the two terms describe different steps. Scraping is the collection of content from websites, while training is the process of feeding that material into a model. By dropping one of them from its terms, Meta has shifted the vocabulary used to describe its activity without, on its own, changing what that activity is. The update also touches Meta's scraping policy, and readers who depend on the details should check the revised terms directly rather than rely on summaries of them.

The notable omission is robots.txt. That plain text file lets site owners tell automated crawlers which parts of a site they may not visit. It is a voluntary convention rather than a legal mechanism, so it only protects a publisher when the crawler chooses to respect it. Publishers and creators have pressed AI companies to commit to honoring these signals, and Meta's revised language does not include that commitment.

That leaves an open question for anyone who publishes on the web. Meta has adjusted its wording and its policy, but it has not said it will follow the opt-out tools that site owners already use. Whether it adds that commitment, or whether competitors make similar promises, will show how much weight the industry is willing to give to a convention that has never been binding.