Meta Indefinitely Pauses Business Terms That Would Have Let It Scrape Company Websites for AI Training

Meta has backed down. The company confirmed this week that a controversial update to its commercial terms of service, which would have let it scrape the websites of businesses holding Meta Business accounts, "are now paused and will not take effect on November 20, 2026." The announcement replaces a preview link Meta had published alongside the proposed terms, and the company says existing terms remain in effect while it "continues to consult with stakeholders."

The language that started the fight was unusually sweeping. As 404 Media and Nieman Lab detailed in October, the draft update to Meta's Commercial Terms, Product Catalog Terms, Self-Serve Ad Terms and Ad Creative Generative AI Terms would have let Meta "use any content from such domains, web pages, or URLs to provide, develop, improve, research, train, and maintain the integrity of" its products. That reach extended well beyond Instagram and Facebook posts to whatever websites a business account linked out to, including news publishers' own articles.

The three-week retreat that followed was remarkably fast. Nieman Lab first reported the terms on October 15. On October 16, Meta deleted the word "train" from the allowed use cases and narrowed the scope to "advertising and business products." On October 23, it pushed the effective date from October 30 to November 20. On October 28, communications vice president Andy Stone pledged the company would not make policy changes "without a full consultation with industry groups and leaders." The consultation never finished; instead the whole thing is now shelved indefinitely, with no new date and no revised text.

Legal and regulatory pressure did the rest. A lawsuit backed by the Norwegian Media Businesses' Association and 180 Norwegian media companies won a temporary injunction from a Norwegian court suspending the terms. Courts in Brazil issued their own preliminary injunction. In the UK, the Competition and Markets Authority opened a formal investigation into whether Meta was "abusing a dominant position by imposing exploitative terms on publishers and advertisers." In the US, Senators Adam Schiff and Marsha Blackburn wrote to Mark Zuckerberg asking him to withdraw the terms entirely.

A Meta spokesperson confirmed the pause: "We're pausing the changes as we continue to engage with stakeholders and take our time to get this right." For now, that means the broad scraping-for-training language never actually went live, and business accounts' off-platform content is governed by the existing terms while Meta decides what, if anything, to propose next.

The episode is a significant marker in the global fight over AI training data. Within three weeks, a routine-sounding terms update drew court injunctions on two continents, a formal competition investigation and a bipartisan Senate letter. Publishers treated it as an existential issue because the terms would have converted their publicly linked content into raw material for Meta's models without negotiation or payment, and the company's own retreat, first softening the language and then pausing it outright, is an implicit acknowledgment that the original draft went too far.

The underlying question is still unanswered: how much of the open web can a platform claim the right to scrape simply because a business holds an account with it. Meta has not said what a revised version would look like or when one might appear. But every other platform watching the injunctions land will now think twice before writing similar terms, and publishers, for once, won a retreat rather than merely a promise to review.