Meta pauses AI training terms indefinitely after Brazil and Norway injunctions
Meta has indefinitely paused a planned update to its commercial terms that would have let it crawl external websites linked from Facebook pages for AI training. The November 5 notice follows preliminary injunctions in Brazil and Norway, a formal UK competition investigation opened October 30, and a letter from two US senators. Paused is not withdrawn.
The News
Meta has indefinitely paused a planned update to its commercial terms that would have required Meta Business account holders to let the company crawl external websites linked from their Facebook pages for AI training. Meta posted a notice in its Business Help Center on November 5 saying the updates "are now paused and will not take effect on November 20, 2026"; the preview page for the proposed terms was removed.
The pause follows weeks of publisher backlash. Courts in Brazil and Norway issued preliminary injunctions — court orders temporarily blocking the terms — after media organisations sued. The UK Competition and Markets Authority opened a formal investigation on October 30 into whether the terms abuse a dominant market position. And US Senators Adam Schiff and Marsha Blackburn wrote to chief executive Mark Zuckerberg asking him to withdraw the terms. The pressure came from three directions at once: courts, a competition regulator, and Congress.
The terms had already been softened once. Meta removed the word "train" on October 16 and limited the use to "advertising and business products", and the effective date had slid from October 30 to November 20 before the pause. But "paused" is not "withdrawn": Meta says it is still consulting stakeholders, and critics note the revised terms never explicitly confirmed respect for robots.txt preferences — the file websites use to tell crawlers what not to copy.
"Paused" is not "withdrawn" — Meta says it is still consulting stakeholders.
Why It Matters
For Meta, training data is fuel. AI models are built by studying vast amounts of text, images and other content, and the terms would have given Meta a huge new source: every website linked from a Facebook business page. That is why the company tried to write the right into its commercial terms in the first place — it was a cheap, broad route to more data at a moment when every AI lab is hunting for it.
For publishers, the fight is about control and payment. Their content would have trained Meta's AI without a deal, and the backlash shows regulators are willing to treat platform terms as competition issues, not just copyright ones. A preliminary injunction in Brazil and another in Norway, a formal UK investigation into abuse of a dominant position, and a Senate letter asking for withdrawal — that is three different kinds of authority pushing the same way, and it is why a company that had already softened the terms once decided to pause them altogether.
There is also a timing angle. The terms had already slipped once, from October 30 to November 20, and had already been softened — the word "train" removed, the use narrowed to advertising and business products. Each retreat bought Meta time but also signalled that the company knew the terms would not survive contact with regulators and publishers. The indefinite pause is the same move, bigger: rather than fight injunctions in Brazil and Norway and an investigation in Britain while senators in Washington asked for withdrawal, Meta chose to stop the clock. The question now is what it does with the time.
Analyst's Take
Our call is Mixed. As a regulatory retreat, the pause caps a legal overhang: injunctions in two countries, a formal UK competition investigation, and a bipartisan letter from US senators were all converging on the same terms, and a company does not want to fight that on three fronts at once. Stepping back removes a source of fines, lawsuits and bad headlines.
But the same pause is a constraint on Meta's AI training-data ambitions. Training data is the raw material of AI models, and the terms would have opened a huge new source — every website linked from a Facebook business page. Losing that route, even temporarily, narrows how Meta can feed its models. The market's calm reaction — META closed Friday, November 6 at $610.15, up 0.84% — says investors see the legal relief first. The longer question is whether Meta can find the data another way.
Are you bullish or bearish on this story?
By The Numbers
FAQ
What would the terms have done?
They would have required Meta Business account holders to let Meta crawl — automatically read and copy — external websites linked from their Facebook pages, so the content could be used in Meta's AI systems. Businesses that refused would have had to stop using Meta Business accounts.
What is a preliminary injunction?
A preliminary injunction is a court order that temporarily blocks something while a case is still being decided. Courts in Brazil and Norway issued them against Meta's terms after media organisations sued — meaning the terms could not take effect there while the lawsuits play out.
What is robots.txt?
It is a small file a website owner puts on their site to tell automated crawlers which pages they may and may not copy. Critics note Meta's revised terms never explicitly confirmed they would respect robots.txt preferences — a key demand from publishers who use it to keep AI crawlers out.
Is the plan dead?
No. Meta says the updates are paused, not withdrawn, and that it is still consulting stakeholders. The terms could return in revised form. What has changed is the timetable: the November 20, 2026 effective date is off, and the preview page for the proposed terms has been removed.
Why does training data matter so much to Meta?
AI models are trained on vast amounts of text, images and other content — the data is the raw material. The more high-quality content Meta can legally use, the better its models. Losing a route to a huge new source of web content constrains how fast and how cheaply it can improve its AI.
👁
Source: Reuters