Meta is pushing back against an interpretation of revised commercial terms that some legal experts said could have enabled the company to scrape news websites for artificial-intelligence training. The company says its crawler behavior did not change, while an independent report says the disputed language later disappeared from the current terms page. The record is therefore a versioned terms controversy—not proof that Meta began scraping, stopped scraping, or received a definitive legal ruling.

Meta rebuts AI-scraping interpretation as disputed terms language disappears

What the terms dispute is about

The episode began with Meta’s commercial-terms update for businesses and other commercial users. Meta’s communications vice president, Andy Stone, said on October 16 that the change was about consolidating product information covering advertising and recommendations, rather than changing how Meta’s crawlers operate. He also said that robots.txt preferences still apply. Those are Meta’s statements, set out in Stone’s public post, not an independent measurement of the company’s crawler activity. Stone’s statement

The controversy turned on wording in an earlier or preview version of the terms. Legal experts cited by Nieman Lab interpreted that language as potentially giving Meta permission to scrape news sites and use material for training. Some also said the wording might conflict with robots.txt, the mechanism site operators use to express automated-access preferences. Those readings are legal interpretations reported by the publication; they are not a court holding, and the supplied evidence contains no forensic crawler evidence. Nieman Lab’s legal-context report

That distinction matters. A contract can describe rights or permissions without demonstrating that a company exercised them. Conversely, the absence of an explicit permission in a displayed version does not by itself prove that no automated collection occurs under another policy, technical pathway, or legal theory. The supplied reporting supports a narrower conclusion: language that prompted concern was reported as present in one version and absent from another, while Meta denied a change in crawler behavior.

This story also differs from broader AI-and-international coverage that focuses on model releases, cross-border policy, or geopolitical competition. Here, the central question is documentary and operational: what did particular versions of Meta’s terms say, what did the company say about its systems, and what can—and cannot—be inferred from a page changing? It is not a report that Meta has launched a new scraping program.

Meta’s response

Meta’s public rebuttal has three parts. First, Stone said there was no change to crawler behavior. Second, he characterized the terms work as a consolidation of information about products used for ads and recommendations. Third, he said robots.txt preferences continue to apply. Taken together, the statement rejects the idea that the terms update itself changed the company’s technical approach to websites. Read the primary statement on X

The wording “no change” should be read as a company assertion about behavior, not as an independently verified benchmark. The supplied evidence does not include crawl logs, a before-and-after technical audit, a list of affected domains, or a forensic examination of Meta’s traffic. It therefore cannot establish whether crawler activity changed in practice. It can accurately report what Meta said it changed—and what Meta said it did not change.

Robots.txt is similarly important but should not be treated as a complete answer to the dispute. Meta’s statement says those preferences apply. The legal-context reporting described earlier indicates that some experts nevertheless considered the disputed terms potentially inconsistent with robots.txt. The two points can coexist as a matter of reporting: Meta states a policy position, while outside experts questioned how contractual language might interact with that preference. Neither side’s interpretation is a judicially settled result in the supplied sources.

The live-page and preview-page discrepancy

The Verge reported on October 17 that the controversial language was no longer on Meta’s current commercial-terms page after the backlash. The current official page is the version presented as the operative terms page at the time of the report. The Verge’s report on the language disappearing Meta’s current commercial terms

But the page history was not uniform. The supplied notes identify a material discrepancy: the current effective page reportedly omitted the crawler text, while Meta’s preview page reportedly still showed the text and carried an October 30, 2026 effective date. Meta’s commercial-terms preview This means a reader could encounter different language depending on which page was viewed. It also means that “Meta removed the language” needs a qualifier: the report concerns the current page, while the preview version reportedly retained it.

The Verge’s October 16 report described the concern that the business terms might permit Meta to scrape material from news publishers. Its follow-up reported the disappearance of the language. These are independent reports about the wording and the public response; they do not independently confirm the crawler behavior that Meta addressed in its rebuttal. The Verge’s initial report

The other approach focuses on the words of the agreement and the breadth of rights they might appear to grant. Nieman Lab reported that legal experts viewed the earlier wording as potentially allowing scraping for AI training and perhaps conflicting with robots.txt. That interpretation treats the contract as significant even without evidence that Meta exercised the alleged permission. It is a forecast of legal meaning, not a finding of fact about a live scraping operation.

For readers, the safest formulation is consequently precise: legal experts said the earlier language might have permitted scraping or training; Meta said crawler behavior did not change and robots.txt still applies; an independent report said the language disappeared from the current page; and no court ruling or forensic crawler evidence is supplied. Anything stronger would turn an attributed interpretation into an unsupported conclusion.

What remains unknown

The first unresolved issue is practical behavior. There is no supplied independent test comparing Meta’s crawler activity before and after the terms update. Without that evidence, the report cannot say that Meta scraped more, scraped less, or did not scrape at all. Meta’s denial is newsworthy, but it remains a statement from the company whose conduct is being discussed.

Why this is not Google’s paid-news pilot

The Meta dispute should not be merged with Google’s paid News AI pilot. The supplied research identifies that pilot as a separate, earlier site topic and specifically distinguishes it from this story. Meta’s episode concerns unilateral wording in commercial terms, a company rebuttal about crawler behavior, and a discrepancy between current and preview pages. Google’s pilot, by contrast, is described in the supplied notes only as a paid distribution experiment; no further details are needed here.

The distinction changes the question being asked. In a paid pilot, the focus is whether publishers opt in to a distribution arrangement and what the commercial exchange is. In Meta’s terms controversy, the focus is whether contract language could be read as permission for collection, how that language relates to robots.txt, and what technical behavior did or did not change. Treating the two as equivalent would imply facts not supplied about participation, payment, consent, or implementation.

FAQs

Did Meta admit that it scraped news sites?
No. The supplied evidence says Meta’s Andy Stone stated that crawler behavior did not change. Independent reporting described disputed language and its disappearance from a current page, but no forensic crawler evidence is supplied.

Did Meta permanently delete the language?
That cannot be established. The Verge reported that the language was absent from the current terms page, while the preview page reportedly still showed it with an October 30, 2026 effective date.

Did a court rule that Meta could scrape or train on news content?
No court ruling is supplied. Legal experts quoted in contextual reporting interpreted earlier wording as potentially allowing scraping or training; those interpretations remain attributed and unsettled.

Did the terms update change Meta’s crawler behavior?
Meta said it did not. The supplied sources do not include an independent before-and-after crawl study, so actual technical behavior remains unresolved.

Does robots.txt settle the issue?
Meta said robots.txt preferences still apply. Legal experts nevertheless debated how the earlier wording might interact with those preferences. The sources do not establish a final legal answer.

Is this the same as Google’s paid-news pilot?
No. The supplied research treats Google’s pilot as a distinct experiment, while this story concerns Meta’s commercial terms, crawler statement, and page-version discrepancy.

References

This article reports the supplied evidence and is not investment or legal advice.