Meta Walks Back Terms Language That Suggested News-Site Scraping for AI

Meta is updating its Commercial Terms again to remove language that suggested it could scrape content from external sites, including news sites, to train its AI. The Verge reported the reversal on Oct. 17. The edit is narrow in wording but consequential for publishers tracking training-data permissions.
The clause at issue said Meta may use any content from domains, web pages, or URLs to provide, develop, improve, research, train, and maintain Meta products. That text no longer appears on Meta's commercial terms page. Its absence is the operative change.
Meta spokesperson Daniel Roberts said the consolidated crawling terms only applied to ads and business products. He said the terms update does not change how Meta uses web crawlers and does not give Meta new rights that supersede content owners' robots.txt preferences. The distinction matters. Crawler behavior is governed in practice by a mix of crawler code, robots.txt signals, and any separate contractual permission.
Roberts had previously pushed back on reporting about the business-terms update. The Verge noted that pushback on Oct. 16, after Nieman Lab detailed the draft language.
That earlier draft had alarmed publishers. The revision scheduled to take effect Oct. 30 stated that continued use of Facebook, Instagram, or WhatsApp would constitute a business account's consent to Meta scraping its website. Legal experts said the draft would allow the scraping of news sites, according to parallel coverage from Nieman Lab and Editor and Publisher. For news organizations that maintain both a website and a business presence on Meta platforms, that linkage was the core concern.
Meta states the updates to its Commercial Terms will go into effect on Oct. 30, 2026. Meta defines the scope as access or use of Meta Products for a business or commercial purpose, including using ads, selling products, developing apps, managing a Page, managing a Group for business purposes, or using measurement services. The current Licenses section frames the license around content shared, posted, or uploaded on or in connection with Meta Products. For U.S. residents or businesses, most disputes must be resolved by binding arbitration on an individual basis, with class actions and jury trials not permitted.
The language now being removed had blurred two different permission models. One is the platform license, where a business uploads content to Meta services under defined terms. The other is web-wide collection, where a crawler fetches third-party domains subject to technical controls and copyright law. Conflating them in commercial terms created uncertainty for engineering and legal teams responsible for data governance, model provenance, and publisher relations. Clear separation is easier to audit.
The broader context here is worth flagging for practitioners. Terms of service have become a secondary front in AI data sourcing, alongside robots.txt, opt-out headers, and licensing negotiations. Contract wording can be read expansively even when a company says no operational change was intended. That is why outside counsel reacted strongly to the consent-by-continued-use formulation.
In this author's view, the reversal reduces ambiguity without resolving the underlying tension. Publishers still need machine-readable and legally durable ways to express crawling and training preferences. Model builders still need documented, revocable permissions that survive scrutiny. A short clause cannot settle that. Precise terms, tied to observable crawler behavior, can.
For teams building or buying AI systems, the practical takeaway is procedural. Read platform terms as versioned infrastructure. Track effective dates. Keep robots.txt, crawler logs, and contract records in the same review. My kids grew up assuming anything posted online would be copied endlessly, and they were largely right about distribution. Training data is where that assumption now gets negotiated line by line.


