Meta has removed the word “train” from a clause in its updated commercial terms after legal experts and publishers objected to language that appeared to let the company scrape websites linked to business accounts and use the content to train its products. The edit was made without an announcement, and the broader permission to crawl those sites remains in place.
The updated terms, which apply to businesses using Facebook, Instagram and WhatsApp for commercial purposes, were scheduled to take effect on October 30, 2026. Nieman Lab, which first drew attention to the clause, reported that the change appeared on Meta’s terms page between October 16 and October 17.
What the original clause said
According to Nieman Lab’s reporting, the original text stated that a business granted Meta permission to use automated software, including web scrapers, bots, spiders, crawlers and user agents, and to create test credentials, in order to access domains, web pages or URLs associated with the business and retrieve updated content from them periodically.
A following sentence said Meta could use any content from those domains to provide, develop, improve, research, train and maintain the integrity of its products. The terms offered no way to opt out of that permission while continuing to use Meta’s products commercially: businesses that disagreed were told to stop using Meta’s products for commercial purposes.
The clause did not use the words “artificial intelligence.” The concern raised by lawyers was that “train,” read in the context of Meta’s products, would plausibly cover training AI models on publishers’ content.
How legal experts read it
Nieman Lab asked three legal experts to review the language. Matt Topic, a media and intellectual property lawyer, described it as an unusual provision buried in terms of service, noting that it would give Meta access to material that could be worth large sums in the content licensing market. He said it was unclear whether a court would enforce the clause, but that publishers should assume it would be enforceable.
James Grimmelmann, a professor of digital and information law at Cornell, said the plain reading for news publishers was that Meta would be permitted to train AI on any content from their websites, and potentially to use that content in ways that compete with the publishers themselves. A publishers’ trade group described the original terms as outrageous, according to the reporting.
What changed, and what did not
The revised text, as reported on October 17, says Meta may crawl associated domains and periodically retrieve updated information from them to provide, develop, improve, research and maintain its advertising and commercial products. The word “train” is gone, and the stated purpose is narrowed to advertising and commercial products.
Several elements survive the edit. The permission to access linked sites with scrapers and bots, the right to create test credentials, and the right to retrieve content periodically all remain. The October 30 effective date is unchanged.
Meta’s position
Meta says the update does not change how it crawls the web. Spokesperson Daniel Roberts said the changes do not alter the company’s existing crawling practices and that website owners can continue to express their preferences through robots.txt directives, the standard file sites use to tell automated crawlers which pages they may visit. Andy Stone, Meta’s vice president of communications, said the purpose of the update was not to scrape news websites but to obtain accurate product information for advertising and content recommendations.
Why publishers are watching
The dispute lands in the middle of a larger fight over the value of web content as training data. Large publishers have been negotiating paid licensing deals with AI developers, and lawyers quoted by Nieman Lab said the content covered by the original clause could carry a licensing value in the millions of dollars. Under the original wording, much of that material would have been available to Meta at no cost, provided the website was connected to a Meta business account.
For any business that links its website to a Facebook page or other Meta business account, the practical takeaways are to read the new commercial terms before they take effect and to check the site’s robots.txt configuration, since Meta says that file remains the mechanism for signaling crawling preferences. Whether the narrower wording and the robots.txt assurance fully address publishers’ concerns will depend on how Meta applies the terms once they are in force.
