Meta Gets Backlash Over Terms Allowing AI Scraping of News Sites, Clarifies Rules Have Not Changed

Meta has removed wording from its commercial terms that raised concerns about the company scraping news websites and using their content to train AI models. The change comes after legal experts and media industry representatives warned that the updated terms could give Meta broad permission to collect content from websites linked to business accounts, even if publishers had not agreed to license their work for AI training.

It all began after Meta updated its commercial terms of use for businesses, including those that use its platforms for advertising and marketing. The revised document included language allowing the company to use content from domains, webpages and URLs associated with a business account to develop, improve, research and train its products and services.

The wording quickly caught the attention of publishers, many of whom maintain Facebook and Instagram business accounts to reach readers and promote their websites. However, Meta has now clarified that the terms apply to its advertising and business products, and says the update does not change how its web crawlers operate.

Why Did Meta's New Terms Raise Concerns?

The original wording said that Meta could use content from websites associated with a business account to provide, develop, improve, research, train and maintain its products and services. This raised concerns because the language appeared to extend beyond content that businesses directly post on Facebook or Instagram.

In theory, the provision could cover content published on a news website linked to a media company's business account. Legal experts who spoke to Nieman Lab warned that the wording was broad enough to raise questions about whether Meta could use such content for AI training without separate licensing agreements.

Media industry representatives also criticised the change. Jason Kint, who leads a trade group representing digital media publishers, argued that the terms appeared to give Meta rights over content published across the wider web, rather than just material shared on its own platforms.

Theo Bamber, CEO of a UK news media trade group, also raised concerns about the potential impact on the market for licensing content. Publishers have increasingly sought payment from AI companies that use their articles and other material to develop AI models.

The concern was not simply about Meta collecting publicly available information. It was also about whether accepting the commercial terms by continuing to use Meta's business products could be interpreted as granting the company broader rights over content hosted on third-party websites.

Meta Says It Has Not Changed How Its Web Crawlers Work

Following the criticism, a report in The Verge said that Meta revised the wording in its commercial terms. The section that raised concerns no longer includes the word "train", and the company has clarified that the relevant crawling provisions apply to its advertising and business products.

In a statement, a Meta spokesperson said: "We've heard you and want to clarify: the crawling terms we consolidated only applied to our ads and business products, and we've clarified that in the terms themselves. It does not change how Meta uses web crawlers."

Meta also said website owners can continue using robots.txt files to communicate their preferences about automated crawlers. These files allow website operators to specify which automated systems can access particular parts of their sites, subject to how individual crawlers interpret and follow those instructions.

The company stressed that the updated terms do not give it new rights that override those preferences.

Meta Muse Raising Questions About Personal Data

Recently, Meta’s AI assistant, Muse, raised privacy concerns over the amount of information it may have about people, including those who have never directly used the assistant.

As reported earlier, Meta Muse may have a detailed page about a person even if they have never interacted with it. This has prompted questions about what information Meta's AI systems can access, how that information is gathered and how users can understand or control its use.

You can read more about this here: Meta Muse May Have a Detailed Page on You, Even If You Never Used the AI Assistant.

The two developments concern different aspects of Meta's use of data, and there is no evidence in the information available that they are directly connected. However, both underline a growing concern for users and publishers: how much information technology companies can collect and use to develop their products, and how clearly they explain those practices.