Meta drops AI scraping language after publisher backlash

Meta has removed a section from its commercial terms that previously suggested it could use content from any website, including news sites, to train and improve its AI systems. This update comes after criticism from publishers and industry groups who saw the language as a threat to their content control.

Why it matters

The change follows a period of backlash. According to The Verge, Meta’s earlier terms stated it "may use any content from such domains, web pages, or URLs to provide, develop, improve" its services, raising alarms about large-scale scraping for AI training. News publishers and content owners pushed back, arguing this could undermine their copyright and business models. The company has now removed this wording from its published terms.

A Meta spokesperson told TechCrunch that the update does not reflect a change in how Meta actually crawls the web, but rather clarifies its policies. That claim, however, does not address the core concern from publishers: trust. Content owners have grown wary of broad terms that could justify mass data collection for generative AI models. Meta’s decision to change the language suggests it is responding to those legal and reputational risks.

Meta’s terms and the scraping debate

Tech companies have been under scrutiny for using public web data to train AI models. Meta’s earlier terms appeared to give it wide latitude to collect and use any content it could access, not just its own platforms. This included news articles, images, and other material that publishers often guard closely. The specific language removed stated Meta could use content from any domain, web page, or URL, which was seen by some as an open door to wholesale scraping.

According to The Verge, Meta’s removal of the clause came after publishers raised objections and public discussion intensified around AI training practices. While Meta insists its actual data collection practices have not changed, the incident highlights how much legal and industry pressure is now focused on AI training data sources. Similar controversies have affected other AI vendors, leading to lawsuits and content bans. In Meta’s case, this is not the first time its approach to AI content and moderation has drawn criticism. For example, its treatment of AI-generated adult content has also been challenged, as covered in our earlier reporting.

What this means if you run a business

If you own or operate a business with a website, this update means Meta’s terms no longer explicitly claim the right to use your content for AI training. This does not guarantee your content will not be scraped, but it removes a legal justification Meta could point to in the event of a dispute. For publishers, especially in news and media, the change is a signal that large platforms can be pressured to narrow their claims over third-party content.

However, the underlying risk remains. AI companies are seeking data wherever they can find it, and the rules are still being tested in court. If your business model relies on control over your web content, whether for subscriptions, ads, or brand protection, you will need to keep monitoring how platforms update their policies and what they reveal about data use. Tracking changes to commercial terms is now part of the operational risk for many companies, not just legal teams.

This episode also shows that industry pushback can force Big Tech to retreat from broad data claims, at least in public-facing documents. It is worth reviewing your own terms of service and robots.txt settings to see what you are currently permitting. As the AI training debate heats up, clear documentation and explicit boundaries are now a necessity for any business with valuable online content.