Meta AI news site scraping remains a live concern for publishers after the company removed an explicit reference to training from commercial terms due to take effect October 30. Legal experts say the revised wording may still give Meta room to argue that it can collect content from websites connected to Meta business accounts and use it to develop AI-related products. Meta disputes that reading.
The earlier version authorized Meta and parties acting for it to use automated tools, including crawlers and bots, to access, index, collect and retrieve material from domains, pages and URLs associated with a business account, Nieman Lab reported. It said Meta could use that material to provide, develop, improve, research and train its products.
Those terms can apply to accounts used to manage Facebook pages, Messenger accounts, Instagram profiles or WhatsApp channels. The scope is broader than publishers, but news organizations commonly use such accounts, placing their sites within the disputed language, according to Nieman Lab.
Can Meta still use news sites to train AI?
That is unresolved. The Verge reported that Meta removed the word “train” after backlash from the publishing industry. But Matt Topic, a media and intellectual-property lawyer at Loevy & Loevy, told Nieman Lab that the revised terms still permit Meta to use scraped material to develop any “business product.” He said “develop” could be read broadly enough to include training, while noting that the term does not appear to be defined.
James Grimmelmann, a Cornell Law School professor of digital and information law, told Nieman Lab that the remaining language covering sites “associated with” a user’s products or services does not limit Meta from scraping entire sites. Those are legal interpretations, not court findings. The research available does not include the revised terms in full or a judicial ruling on either version.
What Meta says about robots.txt
Meta spokesperson Daniel Roberts told The Verge that the update is not about scraping news sites and does not grant Meta new rights that override website owners’ preferences. He said content owners can use robots.txt to signal those preferences.
Robots.txt is a file through which a site tells automated bots which sections they may or may not crawl. Meta has published instructions for blocking its crawlers through that file, Nieman Lab reported.
The fight is over whether the commercial terms could outweigh those instructions for a customer that continues to use Meta’s business products. Lawyers who spoke with Nieman Lab said the original wording could give Meta an argument that its contract supersedes other blocking directions. Topic said it was unclear whether a court would enforce the provision.
Another unanswered question concerns the original authorization to create “test credentials.” Grimmelmann said this could technically permit Meta to register as a false user and get past registration walls. Meta did not answer Nieman Lab’s questions about whether those credentials could bypass paywalls or other paid-content barriers.
For now, the confirmed change is narrower than either side’s broadest interpretation: Meta deleted an express training reference, while retaining language that lawyers say could still be tested in a future dispute. Meta says it has not acquired a new right to disregard publishers’ crawler controls.
This story draws on original reporting from The Verge.
