Meta's revised business terms authorize scraping of publisher websites

Meta's edit on October 17 deleted a single word, 'train,' and swapped a broad 'all Meta products' scope for the narrower 'advertising and business products.' ^[1]But the operative clause - the one granting Meta the right to use scrapers, bots, spiders, crawlers, and even manufactured test credentials to access, index, cache, analyze, collect, retrieve, and crawl any domain, web page, or URL tied to a business account - was left completely intact. ^[2]That's the part doing the real work in the contract. As Gizmodo's sources put it, the revision 'just limits what Meta is allowed to use publisher content for, but not their ability to crawl and get the content in the first place.' ^[3]News/Media Alliance spokesperson Sam Quigley made the same point about the missing word itself: having removed 'training' from the listed uses doesn't mean it's 'expressly ruled out,' so it's 'still potentially on the table.' ^[3]In other words, Meta narrowed its stated intentions while keeping its technical and legal access exactly where it was - a distinction that matters far more to a contract lawyer than it does to a press release.

Underneath the semantics is a straightforward incentive. Legal observers argue the terms let Meta acquire training-grade content without paying the licensing fees that would otherwise apply - fees attorney Matt Topic estimated could run into 'many millions of dollars' in a proper licensing market. ^[4]That's a notable contrast with Meta's own practice elsewhere: the company already maintains paid AI licensing deals with CNN, Fox News, USA Today, and People Inc. ^[5]A terms-of-service clause that quietly extends similar access to every other business account - without a check, a negotiation, or a payment - is the cheapest possible version of a licensing deal for Meta, and the most expensive possible version for the content owner on the other side.

The terms apply automatically to business accounts and come with no opt-out mechanism - accepting them is bundled with continued access to Meta's ad and business products. ^[6]That requirement alone touches a large slice of Meta's business ecosystem: roughly one million businesses currently use Meta's AI agents inside WhatsApp and Messenger. ^[2]That bundling lands very differently depending on who you are. Small, independent businesses and media outlets that depend on Instagram and Facebook ads - among them small Black-owned outlets and creators - have the least realistic ability to walk away, and their main technical defense, a robots.txt file, is a tool many have never configured. ^[7]Large institutional players have more leverage to object loudly: the News/Media Alliance called the terms 'outrageous and probably illegal' and an abuse of market power, ^[8]while the European Publishers Council is pushing the European Commission to scrutinize the change under competition, digital markets, and copyright rules. ^[9]But even that institutional muscle has only won a wording change so far - not a withdrawal of the underlying permission.

Meta has told reporters that website owners can still rely on robots.txt to block its crawlers. Santa Clara University law professor Eric Goldman's reading of the actual contract language is less reassuring: the terms 'don't expressly say that Meta will respect those robots.txt instructions,' and separately, 'the law doesn't necessarily compel anyone to respect robots.txt files' in the first place. ^[2]That leaves the one tool most publishers would reach for to block unwanted scraping resting on a voluntary promise rather than a contractual or legal obligation - precisely the gap Cornell's James Grimmelmann points to when he frames the choice facing publishers as trusting Meta's self-restraint or abandoning its business tools entirely. ^[5]

On X, one alarmed framing argued Meta would use 'associated' websites to train its AI, alongside a separate, more neutral confirmation from The Verge that the disputed language had been removed. Reddit's reaction was louder and more divided, playing out across three separate threads. One widely-upvoted thread dismissed the edit outright as corporate 'PR laundering,' with one comment arguing that 'the wording changes, but the data vacuum never actually turns off,' and another calling it a 'Techno Fascism' move. A separate, smaller thread pushed back in the other direction, with one comment arguing this kind of scraping 'has always been allowed' and comparing it to how search engines already index sites, drawing a rebuttal that querying and scraping-for-AI-training are not the same thing. A third, more personal thread showed what the asymmetry looks like in practice: a small coaching-business owner described wanting to leave Meta's platforms over the new terms but feeling trapped, since doing so would mean losing their following, group, and leads. At least one dedicated AI-news YouTube channel picked up the story as well, quoting the disputed clause's language directly rather than taking Meta's 'just a clarification' framing at face value. ^[1]