Meta may use broad new commercial TOS to avoid millions in AI licensing fees

Meta simplified its trading TOS. Legal experts warn that this is a power grab from artificial intelligence.

Legal experts warn that Meta will soon update its business terms to require business accounts to allow Meta to train AI on their linked website content, not just their social media posts.

The TOS update, which begins on October 30th, does not allow businesses to opt out, instead suggesting that businesses stop using Meta's services if they do not want to allow Metabots to freely crawl their linked content.

As Nieman Lab reported Thursday, this affects all commercial accounts, but is especially shocking for news publishers who can potentially charge Meta millions in licensing fees to continuously access their articles for AI training. News organizations are currently battling OpenAI over whether such training is fair use without a license, and the meta terms apparently risk granting it broad licenses, regardless of which way the court leans.

"While journalism is not the only industry affected by this update, nearly every news organization has at least one Meta business account, including Neiman Labs," the lab said.

You allow us (or any entity acting on our behalf) to use automated software (such as web scrapers, bots, bots, spiders, crawlers and user agents) and generate test credentials to access, index, cache, analyze, collect, retrieve or crawl domains, web pages, or URLs (and any content accessible through them). is) related to your products or services, or made available through your or users' use of Meta Company's products and services, and will receive periodic updates from such domains, web pages, or URLs. Meta may use any content from such domains, web pages or URLs to provide, develop, improve, research, educate and maintain the integrity of Meta's products, services and features offered under these Terms of Business and other applicable supplemental terms.

Notably, no mention of artificial intelligence is made, but it is inferred from a single use of the word "train" and a reference to automated scratching and crawling.

A Meta spokesperson told Nieman Lab that Meta does not intend to use this condition as a legal basis to override the robots.txt guidelines that prevent scraping on news sites. The company still allows news organizations to opt out of scraping and provides guidelines on how to avoid scraping. A spokesperson told Ars: "There is no change in how meta uses web crawlers; any suggestion to the contrary is false."

However, if the courts confirm that training AI on non-news articles is fair use, news organizations question whether the meta position may change, and the condition can be used to freely access content by ignoring robots.txt guidelines.

Three legal experts told Nieman Lab that it's unclear whether a court will enforce those terms, but they could be used to argue that Meta could remove content from news sites and use that content to train the company's AI products." Particularly troubling, James Grimmelman, a professor of digital and information law at Cornell Law School, told Nieman Lab that Meta could "train AI on any content on its sites" and possibly "even remove some of that content." in competition with self-publishers" to deploy on meta platforms.

In addition, Meta recently started charging users for posting links on their platforms, Niemanlab noted. So, the terms seem to give Meta a path to first charge publishers to share links, then use that cash-for-links program to get permission to scrape those links to train AI. Matt Tupik, an intellectual property and media attorney at Loevy & Loevy, told Nieman Lab that "this is a significant provision to bury in the [terms of] service, given that it gives Meta access to material that would likely cost millions of dollars in the licensing market."

Meta announced the update in a blog post, where the company claimed it was simply "optimizing" its trading terms. "Language about crawling was previously scattered across various terms of use, including terms related to the use of generative AI in advertising, and this update consolidates them into one place," a spokesperson said in a statement to Nieman Lab.

"Content owners can still manage access to their content using robots.txt," the spokesperson said.

However, Ars compared the current terms to the updates, which appear to ignore what Meta calls "streaming." The new terms use fresh language and provide broader regulations.

In Meta's current business terms, there's no mention of "experimental evidence" that experts warned Neiman Labs would allow Meta to bypass paywalls and automate scratching. It also does not mention "valid for trial" in the Meta Product Catalog Terms or the Creative Advertising AI Terms, which are other policies that are incorporated into the Business Terms. This is new language that Meta apparently introduced just this month in terms of self-serve advertising.

A spokesperson for Meta did not immediately respond to Ars' request for clarification.

These self-serve advertising terms are for users who want to allow Meta to remove their content in order to create ads. This makes the policy of requiring "trial credentials" to bypass paywalls perhaps less obvious, as long as it's designed for a specific business purpose.

Legal experts told Nieman Lab that the alleged "optimization" of meta is more akin to the antitrust concept of "tying," where businesses force a buyer to buy a second product as a condition of buying another. They likened it to Google's previous practice of requiring websites to participate in general AI reviews if they want their websites to be indexed for public search. (Google has since started allowing opt-outs.)

Previously, there was only mention of crawlers or bots in terms of creative meta AI, and like the new ad policy, that policy had a limited purpose. It specifies that businesses consent to scraping only in relation to ad content, not to improve all meta products:

We or any entity acting on our behalf may access, index, cache, analyze or crawl (including through the use of automated software) the domains, web pages or URLs provided by you for use in connection with your advertising. If you provide URLs in connection with your advertising content, you grant us and any entity acting on our behalf the right to access, index, cache, analyze or crawl the domains, web pages, URLs and any content available through such URLs and periodically update such URLs for your use in connection with To receive your advertisements or for the purposes mentioned in these terms.

On Meta's blog, the company insisted that "these changes will reduce overlap between our contracts and improve the clarity of the terms of your use of our products."

Nieman Lab is now looking to talk to aggrieved news publishers about the situation to find out if the update might cause a broader drop in news organizations' participation on meta platforms. At least one meta user — Steven Beale, founder and editor of The Steampunk Explorer — who depends on Facebook for traffic is considering leaving the platform because of the update. He told Nieman Lab that he knows his website has already been "scratched," but the TOS update seems like "an arrogant, in-your-face move" on the meta side.

"It just makes me angry and I don't want them to have to deal with it," Bill said.

Ashley is a senior policy reporter for Ars Technica, dedicated to tracking the social impacts of emerging policies and new technologies. He is a Chicago-based journalist with 20 years of experience.