The clause, and its removal
Meta has removed a provision from its updated commercial terms of use, TechCrunch reported on 17 October, citing The Verge. The provision said Meta could use any content from a business account’s domains to provide, develop, improve, research and train its products and services.
That is unusually wide language for a commercial agreement. Read literally, it attaches a licence not to what a business posts on Meta’s platforms but to what sits on the business’s own websites — and “domains”, plural, is doing a lot of work in a document a company accepts by running ads.
TechCrunch does not say exactly when the deletion happened, only that the provision has since been removed.
What Meta says it meant
Meta’s position did not change between the two statements, which is the part worth noticing.
Before the removal, a spokesperson said the updated terms did not reflect any change in how Meta crawls the web. After the removal, the company said the crawling terms only applied to its ads and business products, that it had clarified this in the terms, and that the change does not change how Meta uses web crawlers.
So by Meta’s own account nothing was gained and nothing has been given up: the clause did not expand its crawling, and deleting it does not contract it. What changed is what the document says a business has agreed to.
Why publishers read it the other way
The alarm came from lawyers rather than from Meta’s behaviour. Legal experts interviewed by Nieman Lab called it a remarkable provision and said it could let Meta scrape publishers’ websites, including to train AI models.
The mechanism they described matters more than the wording. A news publisher that buys advertising, or runs any Meta business product, is a business with domains. If the terms grant a licence over content on those domains, then the publisher has consented to training use by signing up to an advertising account — a consent obtained in a place nobody negotiates, which is a different thing from a crawler obeying robots.txt.
That is the question the deletion settles for now: whether permission to train can arrive through a commercial contract rather than through a crawl.
What to watch
Whether the clause reappears in another form, and whether the published terms are versioned in a way that makes the change visible to the businesses bound by them. The broader point survives the deletion: terms of service for advertising products are now a place where AI training rights get written down, and they are read by far fewer lawyers than a licensing negotiation would attract.
