Meta Platforms Inc. revised its Commercial Terms, removing the word ‘training’ from its web crawling clause effective October 30, but legal experts said the updated language remains broad enough to permit extensive data collection.
The change followed objections from European and UK publisher groups, but critics argued the revised terms still allow Meta to scrape data widely for various business purposes, potentially including artificial intelligence development.
The updated clause permits web crawling for advertising and business products, and for ensuring the integrity and security of Meta’s products.
Legal experts, however, expressed reservations about the scope of the revised terms. Matt Topic, a partner at Loevy & Loevy, told Nieman Lab that phrases such as ‘develop’ and ‘associated with your products or services’ are sufficiently expansive to cover extensive data scraping, including for AI training.
James Grimmelmann, a professor at Cornell Law School and the University of Wisconsin, echoed these concerns, stating that the new language still grants Meta considerable latitude for data usage.
Daniel Roberts, a spokesperson for Meta, said the update was intended to clarify existing practices and was not related to scraping news sites, according to The Verge.
Previously, the European Publishers Council and the UK’s News Media Association had voiced strong objections to the original wording of Meta’s crawling terms, specifically concerning the explicit mention of ‘training’ data.
The new language does not explicitly reference the robots.txt protocol, a standard for website owners to manage crawler access, although Meta claims it respects these preferences.
Meta’s crawler documentation lists separate crawlers designated for AI training and Meta AI search, suggesting that businesses can utilize robots.txt to manage access for each specific crawler.
Benjamin Sobel, a fellow at the University of Wisconsin, highlighted the potential for Meta to interpret its terms broadly given the absence of explicit restrictions on AI training in the revised text.
Source: Search Engine Journal
