Meta Delays Business Terms After Publishers Push Back on AI Scraping

Meta said it will delay the start of updated business terms after publishers across the United States and Europe warned that the new language could allow the company to harvest news content for use in AI training.

At the center of the dispute is a question with no clear legal answer yet: when a publisher uses a company's advertising tools or embeds its technology, does that give the company the right to collect and use the publisher's content to build AI systems? Meta's new terms, read by legal experts, suggested the answer might be yes.

How Meta's Business Terms on AI Scraping Drew Publisher Anger

Meta's updated business terms were set to take effect on October 30. As written, the terms appeared to give the company wide latitude to crawl websites that rely on its platforms, then use that material for its AI products. The issue is especially pointed for news publishers, because many of them use Meta's advertising tools to reach audiences and generate revenue.

After the terms drew criticism, Meta edited them to remove specific language about training its products. The company framed the edit as a clarification. But legal experts who reviewed the revised text said the change was not enough to rule out AI use of scraped content. Removing one word, they argued, does not change what the technology can do with the material it collects.

Meta spokesperson Daniel Roberts told The Verge that the terms will now go into effect on November 20, "to allow more time for us to explain to our partners and stakeholders..." Roberts also said the delay would give Meta a chance to clarify how it uses web crawlers, and to confirm that content owners can still use robots.txt to set their preferences.

Publishers and Advertisers Pushed Back Fast

The response from the publishing industry was coordinated and quick. The News/Media Alliance, a trade group representing news publishers, sent a letter to Meta CEO Mark Zuckerberg on Thursday. The letter demanded that Meta revise the terms and warned that the language as written would cause "severe harm" to publishers.

The pressure extended well beyond the United States. Digital advertisers and publishers based in the European Union and the United Kingdom also raised alarms, urging regulators in their regions to force Meta to halt the new terms before they took effect. That combination of pressure, from American media groups, overseas publishers and European regulators, appears to have pushed Meta toward the delay.

Why AI Scraping Is Such a Sensitive Issue for News Publishers

To understand why publishers reacted so strongly, it helps to know a little about how AI systems are built. Large language models, the technology behind chatbots and AI writing tools, are trained on massive amounts of text. That text has to come from somewhere, and the open web has historically been a major source.

News articles are particularly valuable as training data. They are written by professionals, edited for accuracy and cover a wide range of topics in clear language. For an AI company, access to a large archive of news content can meaningfully improve the quality of its models.

For publishers, this creates a real problem. If their work is used to train AI, the models that result may eventually answer reader questions directly, without directing those readers to the original news site. That cuts into web traffic, which is how many publishers earn advertising revenue. From the industry's view, their content is being used to build a competitor.

The robots.txt file, which Meta's Roberts mentioned in his statement, is one tool publishers use to signal that their content should not be scraped. It is a standard part of how the web works, and most legitimate web crawlers respect it. But it is not a legal document, and its protections depend on voluntary compliance.

What Happens Between Now and November 20

Meta has framed the delay as a communication effort. Roberts indicated the company wants to use the extra time to talk with partners and stakeholders and explain its intentions. But the industry groups that pushed back on the terms have not said they expect the conversations to produce major changes.

The News/Media Alliance and similar organizations have argued that AI companies should not be able to use published work for training without explicit permission from the creators. Removing a specific word from terms of service, in their view, does not settle that question. It simply makes the language less transparent.

Whether Meta revises the terms further, adds new opt-out tools or holds firm on the current framework will become clearer as November 20 approaches. Some publishers may also push for direct licensing agreements, which would let them negotiate payment for the use of their content rather than relying on opt-out systems.

The Larger Fight Over AI and Content Rights

Meta is not alone in facing this kind of pressure. Across the technology industry, AI companies are in disputes with publishers, authors, musicians and visual artists over the use of creative work in training sets. Several major lawsuits are working through the courts, and regulators in Europe and elsewhere have started to ask hard questions about who owns the data that feeds AI systems.

The outcomes of those cases and investigations will shape the rules for years to come. In the meantime, individual contract terms like Meta's business terms become important pressure points. A clause buried in an advertising agreement can have significant practical effects on whether a publisher's work ends up in an AI training set.

For smaller news outlets, the stakes are especially high. Large media companies have legal teams that can scrutinize contract language and negotiate for better terms. Smaller local or independent publishers often do not. If the final terms give Meta room to use scraped content for AI, those publishers may face a simple choice: accept the terms or stop using the tools entirely.

What to Watch

The November 20 enforcement date is the next key marker. Before then, watch for public statements from the News/Media Alliance or other groups about whether their concerns have been addressed. Watch also for any regulatory action in the EU or UK, where regulators have already been asked to intervene.

Over time, court rulings on the legality of training AI on scraped content could reshape this fight. A ruling that gives publishers clear legal rights over how their content is used would shift the balance significantly. So would a ruling that goes the other way.

For now, the delay is a short-term response to organized pressure. It does not resolve the underlying question of whether Meta's business terms, in their current or revised form, give the company rights over publisher content that publishers did not intend to grant.