Meta removes AI training wording from business terms after backlash

Meta revised its commercial terms for business accounts on Oct. 16, removing the word "train" from the permitted uses of content it scrapes from associated websites, according to Nieman Lab. The revision also narrows Meta's permitted use of the content to "advertising and business products," dropping the earlier coverage of every Meta product. The terms take effect Oct. 30, and business accounts have no option to opt out before then.

Meta's Clarification

Meta spokesperson Daniel Roberts said the company wanted to clarify the language. "The crawling terms we consolidated only applied to our ads and business products, and we've clarified that in the terms themselves," Roberts told The Verge. "This is not about scraping news sites, and these updates don't give us new rights that supersede those preferences."

Roberts said publishers and other site owners can register their wishes through robots.txt, a text file that instructs crawlers on the sections of a site open to them, according to The Verge. Meta has left any pledge to honor robots.txt out of the revised terms, according to Nieman Lab.

Legal and Industry Concerns

The change followed a Nieman Lab story published Oct. 15. It reported that the original terms let Meta use content from sites associated with business accounts to train its products, a category that includes "AI at Meta." Three legal experts told Nieman Lab the language could support an argument that Meta may scrape news sites for AI training. Nearly every newsroom maintains at least one business account with Meta, according to Nieman Lab.

Those experts said the revision leaves concerns in place. Matt Topic, a media and intellectual property attorney at Loevy & Loevy, told Nieman Lab that the revised version of the new terms still allows Meta to use that scraped content to develop any "business product." Topic added, "I think 'develop' is broad enough to include 'train.'"

James Grimmelmann, a professor of digital and information law at Cornell Law School, told Nieman Lab the revision "does nothing to restrict" the "associated with your products or services" language "that would let Meta scrape entire sites."

Benjamin Sobel, a University of Wisconsin law professor, told Nieman Lab that compliance with robots.txt is voluntary. Unless a preference is legally enforceable, he said, "it can still be ignored all the same."

Publisher groups have also objected, according to Nieman Lab. The European Publishers Council said it would petition E.U. regulators and the European Commission to scrutinize the terms. The U.K. News Media Association called on government officials to intervene. Danielle Coffey, president and CEO of the News/Media Alliance, called the terms "outrageous and probably illegal" and said they "cannot be allowed to be implemented on October 30."

Ongoing Legal Battles

Meta faces a separate legal fight over AI training. Five publishers and author Scott Turow filed a class-action lawsuit in May alleging the company pirated millions of books and journal articles to train its Llama models. Meta disputed the allegations and said it would fight the suit.

Why it matters

The revision highlights the tension between tech companies' need for data to develop AI models and publishers' efforts to protect their intellectual property. While Meta removed the specific word "train," legal experts argue that broader terms like "develop" may still allow similar practices, leaving the legal status of web scraping for AI training unresolved.

---

Sources

• qz.com: Meta removed AI training wording from business terms after backlash
• gizbot.com: Meta Gets Backlash Over Terms Allowing AI Scraping of News Sites, Clarifies Rules Have Not Changed - Gizbot News
• streamlinefeed.co.ke: Meta removes AI training clause from commercial terms following publisher backlash | Streamline

This article was drafted with AI assistance and checked against the sources above. Company claims are reported as claims. Cover image is AI-generated.