The dispute began with Meta’s planned commercial-terms update, due to take effect on October 30. The language authorized Meta and parties acting for it to crawl domains and URLs tied to business accounts, then use available content to “provide, develop, improve, research, train” and maintain its products.¹

For publishers, the concern was not subtle. The terms appeared to cover websites linked to Facebook, Instagram, WhatsApp and Messenger business accounts — a category that reaches virtually every news organization. Critics argued that, rather than negotiating licences for articles used in AI systems, Meta could make access to its commercial tools conditional on accepting sweeping scraping rights.

Legal specialists interviewed by Nieman Lab said the provision could be used to argue Meta had permission to scrape news sites and train its AI products. Cornell professor James Grimmelmann warned that Meta could “train AI on any content from their sites” and potentially deploy material “even in competition with the publishers themselves.”¹ The wording also raised questions over “test credentials,” which critics feared might facilitate access behind paywalls.

Meta pushed back, saying the update was an administrative consolidation of terms previously scattered across its policies. It said content owners could still use robots.txt to manage access, and that it did not intend to use the terms to override instructions blocking crawlers. Yet the company’s position faced skepticism because the revised provision appeared broader than earlier ad-focused rules.¹

The backlash then forced a visible retreat. Meta removed the disputed provision and said it had “heard” critics: the consolidated crawling terms applied only to ads and business products, it said, and the revised wording did not change how Meta uses web crawlers.²

The episode leaves the larger fight unresolved. Meta says it clarified existing practice; publishers saw a warning that platform terms can quickly become a bargaining tool in the battle over who pays for AI training data.