Meta's New Commercial Terms Open the Door to Mass Scraping of News Publisher Sites
Meta announced last week an update to its Commercial Terms taking effect on October 30, 2026, with implications that could reshape the business model of journalistic publishing worldwide. The new rules authorize the company to use scrapers, bots, and other automated agents to access, index, cache, analyze, collect, and continuously crawl any domain or web page "associated with" business accounts on Meta's platforms — which includes news publisher sites that operate Facebook Pages, Instagram business accounts, or WhatsApp Business profiles.
The update offers no opt-out option. The terms read plainly: "If you do not agree to the updated terms, please stop all access or use of our Products for business or commercial purposes." For publications that depend on organic Facebook and WhatsApp traffic for content distribution, the alternative between accepting scraping or losing visibility on Meta's platforms is not a choice — it is an ultimatum.
What the new terms permit
Under the revised Commercial Terms, Meta authorizes the use of automated software — including "web scrapers, bots, robots, spiders, crawlers, and user agents" — to access and periodically pull updates from domains, web pages, and URLs "associated with your products or services" or "made available through your or users' use of Meta Company Products and services," as described by legal experts consulted by the Nieman Journalism Lab.
The crucial point lies in the final sentence of the relevant passage: Meta may use "any content" from these domains to "provide, develop, improve, research, train, and maintain the integrity" of its products, services, and features — including artificial intelligence services. This means articles published on a newspaper's own website, not just posts shared on Facebook, can be collected and used to train models like Llama and Meta AI.
The scraping is not limited to a single capture. The terms authorize Meta to "periodically pull updates" from these domains, meaning continuous crawling — precisely the kind of behavior that news sites update multiple times per day.
Why this matters for journalism
Journalistic content has traditionally been considered one of the highest-quality datasets available on the internet: it is reviewed, edited, factual, and structured for human reading. For language model trainers, news articles are gold — exactly because they are produced by professionals and pass editorial filters before publication. The difference from personal blogs or social media posts is not just quality; it is market value.
Several AI companies have already signed licensing agreements with publishers to obtain legal access to that content. News Corp, for example, signed a multiyear deal with Meta valued at up to $50 million per year covering The Wall Street Journal and other group brands. Wikipedia, through Wikimedia Enterprise, has paid agreements with Meta, Microsoft, Amazon, Perplexity, and Mistral AI. But these are bilateral negotiations between publishers with significant bargaining power.
Meta's strategy, as reported by Ars Technica, appears to be bypassing that need for individual negotiation. Instead of sitting down with thousands of smaller and mid-sized publications to negotiate licenses one by one, the company is trying to solve the problem contractually: simply include scraping authorization in the Commercial Terms, and all business accounts that continue using Meta's platforms will automatically consent to content collection.
Some analysts have called it an "in-your-face move" — a power grab that substitutes contractual imposition for negotiated licensing. If upheld, the model would eliminate the need to pay millions in licensing fees to hundreds or thousands of smaller publications that lack the bargaining power of News Corp or The New York Times.
The legal argument and Meta's risks
Three digital law experts consulted by Nieman Lab noted that it is unclear whether a court would uphold this interpretation of the terms. Contract law in many jurisdictions requires that clauses granting significant intellectual property rights be explicit and negotiable. The question remains whether a unilateral update to Terms of Service, with no opt-out and the threat of platform expulsion, constitutes a valid contract for the assignment of rights over content hosted on third-party domains.
There is also a tension with robots.txt directives. If a news site explicitly blocks crawlers in its robots.txt file — something that has become increasingly common since Cloudflare and other providers began offering tools to block AI bots — can Meta legally ignore that block based on Commercial Terms? The question has not yet been submitted to a court.
The broader legal context also weighs heavily. In May 2026, five major publishers — Elsevier, Cengage, Hachette, Macmillan, and McGraw Hill — filed a class-action lawsuit in Manhattan against Meta, alleging unauthorized use of millions of works to train the Llama model. In October, novelist Christopher Farnsworth joined the suit. In September 2026, the Third Circuit affirmed that Westlaw headnotes are copyrightable and that ROSS Intelligence's use of them to train a competing model does not qualify as fair use. The judicial precedent against unlicensed scraping for AI training is solidifying.
If courts confirm that training on non-news content is fair use, news organizations question whether Meta's own stance might change — but for now, the company is pressing forward with the new policy.
What to expect
October 30 is just days away. Unless there is a coordinated industry response or regulatory intervention — the News/Media Alliance has already issued a public statement on the changes — hundreds of thousands of business accounts on Meta's platforms will be, without explicit consent, authorizing the collection of content from their own websites for AI training. The question is no longer whether Meta can do this technically. It is whether it can do this legally — and whether, in the end, it will pay more for a tortuous path than it would have through direct negotiation.
Sources: Ars Technica, Nieman Journalism Lab, ZDNET
✓ Independent sources cross-checked and verified before publishing
