Meta pulls controversial AI-scraping language from its commercial terms after publisher backlash

What happened

Meta recently updated its commercial terms of use with a clause stating that the company can "use any content from [a business account's] domains, web pages, or URLs to provide, develop, improve, research, train, and maintain the integrity of" Meta products and services. Legal experts interviewed by Nieman Lab described it as a "remarkable provision" that appeared to give Meta permission to scrape the website of any publisher with a Meta business account and use that content to train its AI models.

The timing was sensitive: Meta defeated a high-profile copyright lawsuit last year, but other cases against AI companies—including Meta—remain ongoing. The new language quickly drew scrutiny from publishers and legal observers who feared it expanded Meta's claimed rights over third-party content.

A Meta spokesperson pushed back, insisting the updated terms did not reflect a change in how Meta crawls the web. The Verge subsequently reported that the provision had been removed.

In a statement, Meta 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."

Why it matters

The episode underscores how sensitive AI training practices have become for the publishing and developer ecosystems. Even language that a platform frames as a clarification—not a policy change—can trigger concern when it touches on scraping and model training, because the legal landscape around AI and copyright remains unsettled.

For publishers and businesses that rely on Meta's ads and business tools, the concern is structural: agreeing to a platform's terms in order to use its business products should not implicitly grant broad rights to repurpose their content for AI development. Meta's quick retreat suggests the company recognized the reputational and legal risk of leaving the language in place.

For the broader AI industry, it's another signal that terms-of-service language is becoming a frontline in the debate over training data—alongside lawsuits, licensing deals, and robots.txt disputes.

What to watch

• Whether Meta or other platforms introduce more narrowly scoped terms that explicitly separate ads/business-product permissions from AI training rights.
• How ongoing copyright litigation against AI companies shapes future terms-of-service language across the industry.
• Whether publishers push for standardized opt-out mechanisms that go beyond platform-specific terms.
• Meta's next public communication on its web-crawling and AI training practices, given the gap between how it describes its policies and how external experts interpret its terms.