Meta pauses commercial terms update that would have allowed AI training on linked websites

A pause after weeks of criticism

Meta has paused an update to its business terms that was due to take effect on 20 November 2026, and it has not set a new date. The pause covers four documents: the commercial terms, the product catalogue terms, the self-serve advertising terms and the terms for generative AI in ad creatives. The commercial terms apply to any business using Meta's products, including one that only runs a Facebook Page, which would have widened the reach of the change beyond advertisers. Meta told Reuters that its existing terms stay in effect while it talks with stakeholders.

“We're pausing the changes as we continue to engage with stakeholders and take our time to get this right.”

What the wording could have allowed

The draft language drew criticism for seeming to let Meta crawl external websites linked to business accounts and use the material to train its AI systems. Meta later removed the word 'train' from the terms in an effort to calm users, but Theo Bamber argued that 'develop' posed the same threat. Legal experts warned Nieman Lab that the vague wording could let the company disregard robots.txt preferences and crawl large volumes of off-Facebook content. Ars Technica reported that businesses had no way to opt out, and that keeping a Facebook Page meant accepting crawls of every linked external site. Publishers would also have had to create test credentials so Meta's bots could get past paywalls. Meta has countered that the update only gathered crawling rules already spread across several documents, and that site owners can still block its crawler through robots.txt, a point the terms themselves did not spell out.

Delays, courts and regulators

The effective date was first set for 30 October, then moved to 20 November after pushback from Facebook users, courts and regulators. Brazilian and Norwegian courts granted preliminary injunctions that halted the terms once local groups sued, Nieman Lab reported. The UK's Competition and Markets Authority has opened a formal probe to determine whether Meta is abusing market power to obtain AI training data. The European Commission is assessing a complaint from publishers, and US senators Adam Schiff and Marsha Blackburn wrote to Mark Zuckerberg urging him to pull the terms in their current form. Theo Bamber, chief executive of the UK's News Media Association, called the company's conduct 'staggering' in a LinkedIn post.

“It's saying it can use all the content it scrapes to develop 'any business product'. I mean, come on.”

Where things stand

Meta has not said when the changes will return or whether the crawling rules will be revised. An updated page on its Business Help Centre says the new terms will not take effect on 20 November 2026, and that the company is still consulting. Andy Stone, Meta's head of communications, had earlier said the company would not alter its policies without first consulting groups and industry leaders. The pause currently has no end date, and the existing terms remain in effect.