Meta’s Revised Terms Still Grant Broad Web Scraping Rights

Digital publishers face a widening gap between what Meta says its updated commercial terms allow and what the legal text actually permits. The social media giant removed explicit references to artificial intelligence training from language scheduled to take effect October 30, but the revised contract still grants the company sweeping authority to crawl external websites linked to business accounts on Facebook and Instagram.

The core issue is not whether Meta intends to build large language models using scraped news articles today. It is whether the company has engineered a contractual backdoor that allows it to do so later, or to use crawled content for other commercial purposes, without seeking further permission. Because nearly every digital publisher maintains a business account on Meta’s platforms, the terms effectively function as an industry-wide data access agreement imposed unilaterally.

Meta spokesperson Andy Stone said in a statement that the company had “clarified” the language to reflect that the new terms apply “only to our ads and business products.” He stated the changes “were not about scraping news sites” and that the company wasn’t changing how it uses web crawlers.

The revised section, however, authorizes Meta or any entity acting on its behalf to deploy automated software — including web scrapers, bots, robots, spiders, and crawlers — to access, index, cache, analyze, collect, retrieve, or crawl domains associated with a user’s products or services. According to Nieman Lab, which first flagged the initial update last week, the original draft explicitly permitted Meta to use scraped content to “train” its products. After pushback, Meta replaced that word, restricting the stated purpose to providing, developing, improving, researching, and maintaining advertising and business products.

Legal experts told Nieman Lab that this revision remains a substantial expansion of Meta’s scraping permissions. The ambiguity lies in the phrase “advertising and business products.” If Meta eventually classifies AI-driven ad targeting tools or generative commerce features under that umbrella, the contractual restriction offers publishers no practical shield.

Stone claimed the changes “don’t give us new rights to supersede” content owners’ robots.txt preferences. That assurance carries little legal weight. The robots.txt protocol, developed in 1994, is an industry convention, not a binding statute. AI developers routinely bypass these instructions when ingesting training data, relying instead on fair use arguments or simply ignoring the standard when enforcement mechanisms are absent.

Media organizations view the updated terms through the lens of Meta’s systematic retreat from funding journalism. The company stopped paying for news in 2022 and eliminated its dedicated news tab entirely in 2024. More recently, according to reports, Meta has negotiated a handful of isolated licensing deals with specific outlets solely to supply content for its AI chatbot. This selective approach leaves the vast majority of publishers without compensation while their public-facing websites remain technically accessible to Meta’s crawlers under the new commercial agreement.

Industry leaders have pushed back sharply. Danielle Coffey, CEO of the News/Media Alliance, called the rules “outrageous and probably illegal” in a statement. Theo Bamber, the top executive of the UK’s News Media Association, told Press Gazette the move was a “shameless landgrab attempt” by the company.

Publishers now face a binary choice ahead of the October 30 deadline: accept the expanded crawling permissions as a condition of maintaining their commercial presence on Facebook and Instagram, or sever ties with the platforms that drive significant portions of their referral traffic. Neither option restores leverage over how their external web content is accessed and utilized.