Meta’s updated business terms could let it scrape publisher sites for AI training

Updated October 16, 2026 5:54 pm

Meta’s revised business terms could give the company permission to crawl websites tied to business accounts, including news publishers, and use that material to train and improve its AI systems. The change, which is scheduled to take effect on October 30, raises fresh questions about whether publishers may be forced to choose between broader distribution on Meta’s platforms and tighter control over their content.

The policy shift matters because it could expand Meta’s access to a much larger pool of web content than the licensing deals it has signed so far. Those deals include agreements with several major news organizations, but the new language appears to create a separate route for Meta to collect online material from businesses already using Facebook, Instagram, WhatsApp, and Messenger for promotion.

According to reporting from Nieman Lab, the updated terms could allow Meta to access domains associated with business accounts and use that material for a wide range of AI-related purposes, including development, research, training, and product maintenance. Meta has not publicly commented in detail on the interpretation of the terms, and it did not immediately respond to requests for comment from The Verge.

What Meta’s new terms say

The clearest change is in the language governing commercial accounts. Meta’s terms indicate that businesses authorize the company, or third parties working on its behalf, to use automated tools such as crawlers, bots, and scrapers to access and analyze websites tied to those businesses.

That permission appears to extend beyond a single social post or profile page. The language refers to domains, web pages, and URLs that are associated with a business’s products or services, as well as content made available through the use of Meta’s own products and services. In practical terms, that could mean a publisher’s wider web presence, not just the material it posts directly to Meta’s apps.

Under the updated wording, businesses are effectively granting Meta permission to use automated systems to collect, index, retrieve, and periodically revisit content from related domains, then reuse that content for its products, services, and AI systems.

The policy also says businesses that do not accept the new terms may lose access to Meta’s products for business or commercial purposes. That raises the stakes for publishers and other companies that depend on Meta for audience reach, traffic, or paid promotion.

Why publishers are paying close attention

Publishers are watching closely because the terms appear to create a potential workaround around licensing negotiations. Instead of signing a direct agreement with every newsroom or media company, Meta may be able to rely on the commercial relationship it already has with businesses that advertise, post, or otherwise use its platforms.

That matters in a media environment where AI companies are competing aggressively for high-quality text, and journalism remains one of the most valuable sources of up-to-date information. News reporting is especially attractive because it is timely, structured, and often tied to real-world events that improve the usefulness of generative AI systems.

Nieman Lab argued that the new terms may let Meta “bypass” licensing in some cases by giving it access to material associated with business accounts, including online publishers. The publication also pointed out a growing tension in Meta’s approach: the company has started charging businesses for some forms of link-sharing while potentially reserving the right to crawl those same domains for free.

How does this differ from Meta’s existing AI deals?

It differs because Meta’s licensing deals are voluntary agreements, while the new terms appear to be built into the company’s standard commercial rules. In other words, instead of negotiating access one publisher at a time, Meta may be embedding the right to collect content directly into the terms businesses accept to use its platform features.

Meta has already followed a path similar to OpenAI’s by signing content agreements with a handful of publishers, including CNN, Fox News, USA Today, and People Inc. Those deals are part of a broader industry trend in which AI developers seek formal arrangements with media companies to reduce legal and reputational risk.

But licensing is only one route to training data. If Meta can lawfully use public or associated business websites under its terms, it could access a much broader slice of the web than a small set of news deals would provide.

What changed on October 30?

The new rules are set to take effect on October 30, which gives businesses a short runway to review how the company’s terms affect their content. For some publishers, the timing is especially notable because the media industry is still trying to establish durable rules around AI scraping, attribution, and compensation.

Meta’s change arrives as legal, commercial, and ethical debates over AI training data continue to intensify. Companies across the industry have faced criticism for using publicly available content without clear consent or payment, while publishers argue that their work is being repackaged into AI tools that may reduce the value of original reporting.

In that context, Meta’s revised terms are not just a contractual update. They are another example of how major tech platforms are trying to define the boundaries of data access before regulators or courts settle the issue more definitively.

How big is the AI data appetite at Meta?

Meta’s appetite for training data appears to be growing as the company expands its AI ambitions. The company has made AI a strategic priority across its products, from chat features to content ranking systems and future consumer tools. Large language models generally improve with more and better data, especially when that data is fresh and varied.

That technical reality helps explain why publishers are uneasy. News sites produce a steady stream of topical material, making them especially valuable for models that need to answer questions about current events, public figures, technology trends, and local developments.

It also helps explain why platforms are increasingly trying to formalize data access. When AI systems can be trained on large, diverse corpora, they may become more accurate and more commercially useful. But the same systems can also weaken the economics of publishing if they reduce the need for users to visit the original source.

What does this mean for publishers?

It means publishers may face a difficult trade-off. To keep using Meta’s business tools and reach audiences on its apps, they may have to accept terms that give Meta wide latitude to reuse content associated with their sites.

Some publishers may decide that visibility on Meta is still worth the risk. Others may see the new policy as yet another reason to reduce dependence on large platforms and invest more heavily in direct traffic, subscriptions, newsletters, and owned audience channels.

For smaller outlets, the problem could be even more acute. They may have fewer options for negotiating custom terms or walking away from Meta’s business ecosystem without sacrificing reach.

Why this could set a precedent beyond Meta

This policy update may become a model for other platforms and AI developers looking to expand their training datasets. If the terms stand and businesses accept them at scale, major companies could increasingly wrap AI training rights into the fine print of everyday commercial relationships.

That would be a significant shift in how AI data is sourced. Instead of relying primarily on opt-in licensing deals or publicly accessible content collected under broad web norms, companies could secure access through terms linked to business services, ad tools, and social platform features.

That model could be efficient for tech firms, but it also centralizes power. Businesses may not have meaningful negotiating leverage when the platform controls the audience, the distribution system, and the terms under which content can be reused.

Timeline of the Meta policy change

The current dispute is the latest chapter in a broader series of moves by Meta as it builds out its AI strategy.

1. Last month: Meta introduced a new Meta One subscription that requires businesses and creators to pay if they want to include links in certain posts and reels.
2. Recently: Meta updated its commercial terms to allow crawling of websites associated with business accounts.
3. October 30: The new terms are scheduled to take effect.

The juxtaposition is hard to miss. As Nieman Lab noted, Meta is now charging some users to promote links while reserving the right to crawl related websites for AI purposes. That combination is likely to remain controversial if publishers believe their content is being monetized twice: once through platform access and again through AI training.

What Meta has said so far

So far, Meta has not offered a detailed public explanation of how broadly the new terms will be interpreted in practice. The company also did not immediately respond to a request for comment from The Verge, leaving outside observers to rely on the wording of the policy and the analysis of media watchdogs and legal observers.

That silence leaves room for uncertainty. The exact scope of “associated” websites, the specific rights Meta intends to claim, and the types of content it may prioritize for training are all open questions until the company clarifies enforcement or provides examples.

Still, the language itself is broad enough to draw attention. For publishers already wary of AI companies using web content without compensation, Meta’s new terms may be read as another attempt to expand access to the open web through contract language rather than formal agreements.

What happens next?

For now, publishers and businesses will likely spend the coming days reviewing the updated commercial terms before the October 30 deadline. Some may seek legal guidance. Others may simply continue using Meta’s platforms while weighing the risks in the background.

The larger debate, however, is not just about one policy update. It is about who controls access to the material that powers AI systems, how much consent is truly meaningful when a platform is dominant, and whether the economics of journalism can survive if the web’s most influential companies can convert commercial access into training rights.

Meta’s latest move suggests that the battle over AI training data is moving deeper into the terms and conditions that businesses accept every day. For publishers, that may prove just as consequential as any headline-grabbing lawsuit or licensing deal.

Key facts at a glance

• The updated Meta business terms take effect on October 30.
• The policy appears to cover websites associated with business accounts on Meta’s platforms.
• Meta says it can use automated tools to crawl and analyze that content.
• The company may use the material to develop and train AI products.
• Businesses that reject the terms may lose business or commercial access to Meta products.

As the deadline approaches, the central question is whether publishers and other businesses will accept a contract that potentially opens their web content to Meta’s AI systems — and whether that trade-off will become the new normal across the digital economy.

How could this affect the broader AI industry?

It could encourage other companies to pursue similar terms, especially if the strategy helps them avoid lengthy licensing negotiations. If Meta’s approach holds up commercially and legally, it may show that platform contracts can be as valuable as standalone data deals in building AI models.

That would deepen a trend already reshaping the industry: AI firms want access to more training data, while publishers want compensation, control, and clearer limits. Meta’s revised terms suggest those interests are increasingly being negotiated not only in courts and boardrooms, but in the fine print businesses click through to stay visible online.