Meta Platforms, Inc. (Meta) adjusted its commercial Terms of Service (TOS) on October 16, 2026, by removing the explicit word "train" from a section detailing how content obtained via web crawling could be utilized for artificial intelligence (AI) model training. This change occurred just days before the updated terms were scheduled to take effect on October 30, 2026.

The alteration responds to considerable public and industry backlash, particularly from publishers and legal experts, who voiced concerns that the original language would permit Meta to train its AI models on third-party content without clear consent or compensation. This incident highlights the growing influence of public and industry pressure in shaping the data acquisition policies of major technology companies.

What the records show

The original commercial terms, which were set to become effective on October 30, 2026, had generated widespread alarm. The contentious clause stated that Meta could "use automated software such as web scrapers, bots, robots, spiders, crawlers, and user agents and create test credentials to access, index, cache, analyze, collect, retrieve, or crawl the domains, web pages, or URLs, and any content available through them, that are associated with your products or services". Crucially, this collected content could be used to "provide, develop, improve, research, train, and maintain advertising and business products including business, commercial and related products". Publishers and legal communities raised concerns that this wording would allow Meta to train AI models on external content without explicit permission or payment, potentially bypassing robots.txt preferences. The initial terms also allowed Meta to "periodically pull updates from such domains, web pages, or URLs".

The timeline

Meta’s initial commercial terms were publicized with an effective date of October 30, 2026. Following intense criticism from various groups, Meta announced its revision on October 16, 2026, removing the specific word "train" from the disputed clause.

What changed and for whom

The primary change involves the removal of the word "train" from the clause describing the uses of crawled content. The revised clause now states that Meta's automated data collection aims "to provide, develop, improve, research, and maintain advertising and business products, and to maintain the integrity and security of Meta Company Products". Meta defines "advertising and business products" broadly to include business, commercial, and related products that assist users in creating and delivering ads and other commercial content, provide information about their products and services, and connect them with interested users.

Legal experts and publishers, such as the News/Media Alliance, strongly opposed the original terms, with CEO Danielle Coffey calling them "outrageous and probably illegal". Their concerns encompassed intellectual property rights, data privacy, and the potential override of robots.txt preferences, which are not legally binding. Sam Quigley, a spokesperson for the News/Media Alliance, pointed out that the revision only limits the purpose for which Meta can use content, not its ability to crawl and acquire it. Legal analyses also suggest that terms like "develop" in the updated TOS might still be interpreted broadly enough to include AI training.

What Meta has said publicly

Meta spokesperson Daniel Roberts clarified that the crawling terms were intended to apply "only to our ads and business products" and that the update "does not change how Meta uses web crawlers". Andy Stone, Meta's Vice President of Communications, also stated that the change was "not about scraping news sites" and that "these updates don't give us new rights that supersede those preferences". Both spokespersons consistently framed the revision as a "clarification" rather than a substantive policy change.

What we could not confirm

PrimaryBrief could not confirm whether Meta will explicitly incorporate a legal commitment to honor robots.txt preferences in its terms. We also could not confirm the definitive interpretation of terms like "develop" within the revised TOS, and whether this term might still allow for AI training under a broad reading.