Meta deleted one word from its scraping terms. Publishers say that's not enough

If you run a newsroom, you almost certainly have a Meta business account. It is how you manage your Facebook page, your Instagram profile, your boosted posts. Nieman Lab put it plainly when it broke this story on Oct. 15: virtually every news organization has at least one, including Nieman Lab.

That makes the fine print on those accounts a big deal. And for a couple of days last week, the fine print said Meta could crawl your website and use what it found to train its products.

The sentence that set it off

Meta's new commercial terms take effect on Oct. 30. As TechCrunch quoted them, they gave the company the right to "use any content from [a business account's] domains, web pages, or URLs to provide, develop, improve, research, train, and maintain the integrity of" Meta products and services. That means your own website, beyond anything you post to Facebook.

Nieman Lab reporters Andrew Deck and Hanaa' Tameez asked three legal experts to review the terms. Their verdict, per Nieman Lab, was that as written the terms could be used to argue Meta can scrape content from news sites and use it to train the company's AI products. The terms cover business accounts that manage pages and groups, run ads, send marketing messages or use Meta's business dashboard. Consent works the way it always does with platforms: keep using the tools after Oct. 30 and you have agreed.

This landed in the middle of a fight that publishers already feel they are losing. AI companies want news archives because they are well written, fact-checked and constantly refreshed. Publishers want to be paid for that, and some have signed licensing deals. A clause that quietly turns a Facebook page into a training-data permission slip makes those deals worth less, because why pay for what you already got in the terms of service?

The backlash was fast. Gizbot reported that Jason Kint, who leads a trade group for digital publishers, argued the terms appeared to give Meta rights over content across the wider web, and that Theo Bamber, CEO of a UK news media trade group, warned about the effect on the content-licensing market.

Meta's edit

Within days Meta changed the text. According to TechCrunch, the word "train" is gone, and the crawling provisions now say they apply to Meta's advertising and business products. The company's statement:

"We've heard you and want to clarify: the crawling terms we consolidated only applied to our ads and business products, and we've clarified that in the terms themselves. It does not change how Meta uses web crawlers."

Meta has said from the start that the update does not change how it crawls the web. Website owners, it says, can keep using robots.txt, the plain text file that tells crawlers which parts of a site to stay out of, to set their preferences.

So, problem solved? Nieman Lab went back to its legal experts, and the follow-up headline gives away the answer: they warn that the new terms still grant broad scraping permissions, despite the language changes. And the revised terms, according to Nieman Lab and Quartz, still contain no specific pledge to honor robots.txt. Meta says it respects those files. It just hasn't written that promise into the contract that business accounts are agreeing to.

That gap is the whole story. Robots.txt is a convention. It works because crawlers choose to obey it. A contract is what a lawyer waves in court. If the contract says Meta may collect and use content from your site for its business products, and says nothing about your robots.txt, the experts Nieman Lab spoke to worry that wording could be used to argue it overrides whatever your text file says. Whether a court would agree is unsettled. Nobody wants to be the publisher who finds out.

Narrowing the scope to "advertising and business products" helps less than it sounds. Meta's ad products are among the most heavily AI-driven systems on earth. Content used to build better ad targeting or business tools is still content feeding Meta's machine learning, whether or not anyone uses the word train.

There is also no opt-out. A publisher who objects has one option before Oct. 30: stop using Meta's business tools, which for most newsrooms means cutting off a large share of their audience. That is the leverage. Meta doesn't need to win the legal argument. It needs publishers to decide the argument isn't worth losing Facebook over.

The terms go live in 11 days. Meta deleted a word. Everything else stayed in.

Sources: Nieman Lab: Meta's new commercial terms will allow the scraping of news sites, legal experts say, TechCrunch: Meta removes language that raised scraping and AI training concerns, Nieman Lab: Legal experts warn Meta's new terms still grant broad scraping permissions, Quartz: Meta removed AI training wording from business terms after backlash, Gizbot: Meta gets backlash over terms allowing AI scraping of news sites.