Meta Drops “Train” From Business Terms, but the Scraping Dispute Continues
Meta has removed language from its business terms that said the company would “train” its products on content scraped from news sites, according to The Verge. The revised terms for business accounts are scheduled to take effect on October 30.
The earlier wording said Meta “may use any content” from websites associated with a business account to “develop, research” and “train” its products. That phrasing drew criticism after the terms surfaced. The Verge reported on October 16 that the terms appeared to give Meta permission to scrape online publications for free, citing Nieman Lab's analysis. A follow-up on October 17 reported that Meta was updating the terms again to remove language suggesting it could scrape content, including from news sites, to train its AI. Scraping means using automated software, often called a crawler, to copy pages from the web.
The revision did not satisfy everyone. Cornell Law professor James Grimmelmann said the revised terms do nothing to restrict the phrase “associated with your products or services,” which he argued would still let Meta scrape entire sites. Media attorney Matt Topic said the revised terms still allow Meta to use scraped content to develop any “business product,” and that “develop” is broad enough to include “train.”
Meta spokesperson Daniel Roberts said content owners can use robots.txt to signal their preferences. Robots.txt is a plain-text file on a website that tells crawlers which pages they may visit. He said the update “is not about scraping news sites.”
Nieman Lab had gone further on the consent mechanics in earlier reporting. It stated that business accounts continuing to use Facebook, Instagram, or WhatsApp after October 30 would consent to Meta scraping their websites under its updated Commercial Terms.
In my view, the dispute turns on a narrow question of drafting rather than on the headline change. Removing the word “train” changes what the document says on its face. Whether “develop,” combined with “associated with your products or services,” reaches the same practical result is the question Grimmelmann and Topic are answering, and they answer it differently from Meta's spokesperson. Readers who manage crawl policies should read the operative clauses themselves, not the summaries.
The broader context here is the opt-out mechanism. Robots.txt is a voluntary convention. A publisher can disallow specific crawlers, but the convention depends on the crawler honoring it, and it does not in itself govern what a platform does with content it has already been permitted to access under a contract. Meta's emphasis on robots.txt places the burden of signaling preferences on the content owner, which is a different starting point from a default requiring affirmative permission.
The October 30 effective date gives publishers and business account holders a short window to read the text, decide whether to keep using the services, and configure their sites. If the terms take effect as scheduled, the operative language will be the version that matters. The arguments over “develop” and “associated with” will then move from commentary into whatever enforcement or litigation follows. For now, the most concrete development is a removed word and a date on the calendar.
