Businesses, publishers and news organisations should be uneasy about an update to Meta’s Commercial Terms that is set to take effect on October 30. The new legal term lets Meta send its bots and crawlers into any website “associated with” a business account on its platform.
Harvard University’s Nieman Lab, which first reported the update, noted the change allows the social media to engage in broad-scale scraping and the possible use of that content to train its own AI models.
Following the report, the company revised its policy, removing the word “train” and narrowing the permitted uses to “advertising and business products” instead of all its products. It said the first draft had not made clear that robots.txt preferences are respected. Still, the current terms do not include this updated preference.
Big business
About 3.6 billion people use at least one of Meta’s family of apps daily. Data aggregation sites report that over 200 million businesses use Meta’s apps each month. The company’s advertising revenue reached $59.36 billion in the June ending quarter.
Most of the advertisers buy buy space through self-serve tools like Ads Manager and Business Suite. Businesses build and pay for their own campaigns and accept standard terms with a click.
Continued use of this self-serve feature means the account has accepted the new terms, and, according to the reports, business accounts cannot opt out before October 30.
The fine print
The new policy apply to any “business or commercial purpose,” including running a Page, “regardless of the entity type.” A news media organisation posting to Facebook or buying a boosted post qualifies.
Accepting the new terms means Meta, or anyone acting on its behalf, can use “web scrapers, bots, robots, spiders, crawlers.” Meta may then access, index, cache and analyse its web pages and pull updates periodically. Their stated aim for doing is to develop and improve advertising and business products and to keep the platform secure.
The company’s updated terms, removing “train” narrows the language, it will still use this consent to “develop, improve, research” its products. “Advertising and business products” is not defined in the new terms, and the text says nothing about paywalled content.
The web crawling problem
News site owners and publishers use a text file on their website to tell web crawlers and AI bots to not scrape their content.
But, if a publisher signs a formal, written contract with Meta that grants permission to use automated tools on their site, that legal agreement carries more weight.
Meta can legally argue that the signed contract gives them explicit permission, meaning they can completely ignore any “do-not-scrape” flags or technical blocks the publisher puts on the website.
Legal experts, cited in Nieman Lab’s report, note that the terms could supersede the text file that blocks web crawlers. Meta claims nothing has changed.
If that were so, including it into the terms would cost Meta nothing. The terms also state that changes must be in writing and signed by Meta.
Licensing risks
The crawling clause is not the only problem. News organisations warrant that they hold every right needed to license their content to Meta, including to create works. That is hard to promise for wire copies and agency photographs, which newsrooms often use under narrow licences.
A broad indemnity then makes the account holder cover Meta’s losses and legal fees if “you or anyone” brings a claim over its content.
Meta can discontinue any commercial product “without prior notice or liability.” Its own liability is capped at the greater of $100 or what the account has paid in the past twelve months.
A publisher that spends little on ads has almost no recourse. In the US, disputes go to individual arbitration, with copyright and unauthorised-access claims reserved for Northern California courts. Elsewhere, California or Irish courts apply, and EU business users can invoke a dispute mechanism under the Digital Markets Act.
The gap must be closed
A company with this reach should not ask the industry whose work it may crawl to rely on a verbal promise. The public needs to know what Meta’s bots may read, which tools they feed, and whether any of that content ends up in systems beyond advertising.
If Meta means what its spokesperson said, the fix is one sentence: robots.txt preferences will be honoured, and paywalled content is excluded.
Until that amendment appears in the terms, publishers are being asked to sign over access first and trust later. For news organisations, whose websites are the product, that is the wrong way round.