Meta Quietly Removes Controversial Commercial Terms Language That Sparked Publisher Alarm Over AI Scraping

Meta has quietly removed a controversial update to its commercial terms of use that alarmed legal experts and digital publishers, who warned the language effectively granted the social media giant sweeping permission to scrape news websites and train its artificial intelligence models. The swift reversal follows intense scrutiny over how the tech conglomerate handles publisher content through business accounts, a friction point that underscores ongoing legal and ethical battles between media organizations and major artificial intelligence developers.

The controversy erupted after Meta updated its commercial terms of use, introducing a provision stating that the company could use any content originating from a business account’s domains, web pages, or URLs for a multitude of internal purposes. Specifically, the updated terms indicated that Meta could leverage this external web content to provide, develop, improve, research, train, and maintain the integrity of Meta products and services.

When legal experts analyzed the updated language, many raised immediate red flags. According to legal specialists interviewed by Nieman Lab, the wording appeared to be a remarkable provision that could give Meta explicit contractual permission to scrape the website of any news publisher, media outlet, or commercial entity operating a Meta business account. Crucially, experts warned that the harvested data could then be funneled directly into training the company’s expanding suite of artificial intelligence models without separate consent or negotiation.

The inclusion of this language struck a nerve across the media landscape because it arrived amid a volatile backdrop of intellectual property disputes. Major publishers, independent newsrooms, and creators have increasingly pushed back against the unauthorized harvesting of journalistic work to power generative artificial intelligence systems. While Meta successfully defeated a high-profile lawsuit last year when a federal judge sided with the company in a legal battle concerning the training of AI models on copyrighted books, the legal war is far from over. A steady stream of other lawsuits against AI companies, including Meta, continues to wind its way through the courts as publishers seek clarity and compensation for the use of their intellectual property.

As speculation mounted and concerns over the broad reach of the terms grew, Meta initially defended the update. A company spokesperson insisted that the revised terms did not reflect any operational change in the way Meta crawls the web or indexes external pages. However, the pushback proved significant enough to prompt a policy retreat. Industry reporting from The Verge revealed that Meta has officially removed the contentious provision from its commercial terms, pulling back the language that sparked fears of widespread web scraping for AI training.

Following the removal, Meta issued a clarifying statement attempting to quell the anxiety among publishers and business account holders. The company explained that the confusion stemmed from consolidated crawling terms that were never intended to signal an expansion of its data collection practices.

We 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, the company said in its statement, emphasizing that the adjustment does not change how Meta uses web crawlers.

Despite the swift removal of the language and Meta’s assurances, the episode highlights the profound sensitivity surrounding artificial intelligence training data and the uneasy relationship between platform giants and digital publishers. As technology companies continue to refine and scale their generative AI products, the boundaries between advertising services, business accounts, and automated data harvesting remain a critical battleground for copyright holders worldwide.