IA · 18 October 2026 · 4 min read

Meta Backtracks on Web Scraping: Content-Training Clause Removed from Commercial Terms

In brief: Less than forty-eight hours after widespread criticism from publishers and legal scholars, Meta updated its Commercial Terms of Service to remove a controversial clause. The provision appeared to grant the company broad authorization to crawl and use content from any website linked to a business account to train its artificial intelligence models, prompting reassurances that standard robots.txt directives remain respected.

by Team Mocchi's

Menlo Park beats a retreat

The controversial contract revision that seemed poised to give Meta unprecedented access to the web content of millions of businesses lasted less than forty-eight hours. On October 17, the social media giant issued a swift update to its Commercial Terms page, removing language that would have required any organization operating a business account to accept automated crawling of its external websites.

As reported by The Verge, the disputed passage quietly disappeared from the live terms following warnings from news publishers and legal academics that Menlo Park was staging a contractual override to feed its AI models without negotiating formal licensing agreements.

The legal outcry over the single word "train"

The controversy erupted when observers scrutinized an upcoming revision scheduled to take effect on October 30. The drafted terms authorized Meta to deploy automated tools—including crawlers, scrapers, and user agents—to access, index, cache, and retrieve data from any domain or URL associated with a company's commercial profile.

The real trigger was the explicit scope of use: Meta claimed the right to utilize that external content to «provide, develop, improve, research, train, and maintain the integrity of» its products. According to Ars Technica, the inclusion of the word «train» in a take-it-or-leave-it contract raised alarm across digital newsrooms. Legal experts, including Cornell University information law professor James Grimmelmann, warned that the terms could have allowed Meta to argue in court that publishers had contractually waived their rights to block AI data harvesting, even overriding explicit robots.txt instructions.

Rewriting the terms and reaffirming robots.txt

Facing escalating criticism from industry groups and legal observers, Meta leadership initially downplayed the wording as a consolidation of legacy policies. As scrutiny intensified, however, the company opted to strike the entire provision.

According to TechCrunch, Meta spokesperson Daniel Roberts issued a statement explaining the revision: «We’ve heard you and want to clarify: the crawling terms we consolidated only applied to our ads and business products, and we’ve clarified that in the terms themselves. It does not change how Meta uses web crawlers. As always, content owners can use robots.txt to signal their preferences. This is not about scraping news sites, and these updates don’t give us new rights that supersede those preferences».

The removal restores robots.txt as the accepted technical defense line for organizations unwilling to feed their proprietary materials into commercial machine learning pipelines.

Mocchi's take

This rapid reversal highlights the fragile balance between leveraging platform distribution and maintaining control over enterprise data. For businesses that invest heavily in original content, technical documentation, and digital assets, the incident serves as a crucial wake-up call: maintaining a social media presence should never silently compromise digital sovereignty. We strongly advise engineering and marketing teams to audit their robots.txt configurations regularly and review business terms of service, ensuring corporate visibility does not inadvertently turn into unpaid training fodder.

All articles on the Mocchi's blog