Meta took the crawl-and-train sentence off the commercial terms

The sentence the outlets say is gone

The Verge, on October 17, 2026, reports that Meta removed commercial-terms language which legal readers had said could allow scraping of sites, including news sites, to train AI products. The sentence it quotes, and says no longer appears, authorized Meta to use content from domains and URLs associated with a business in order to provide, develop, improve, research, train, and maintain Meta products. TechCrunch and Indian Express report the same removal. Three accounts of a deletion are not a screenshot this desk captured. They are the record that the sentence which yesterday's brief quoted has been taken down.yesterday's commercial-terms brief.

What the critics had said the sentence did

The Verge says Nieman Lab and James Grimmelmann read the wording as a right to train on scraped content for free and to supersede a site's robots.txt instructions. Indian Express reports the same criticism and the same professor. Yesterday's brief filed those readings against Meta's claim that crawlers were unchanged. Today's fact is narrower and newer: the sentence those readings attacked is, by these outlets' accounts, no longer on the page.the privacy desk.

Roberts's statement draws a box around ads and business products

Daniel Roberts's statement to The Verge says the company heard the objection and clarified the terms. The crawling language that was consolidated, he says, applied only to ads and business products. He says the update does not change how Meta uses web crawlers, that robots.txt remains the signal, and that the update is not about scraping news sites and grants no new right over that signal. Andy Stone's shorter line, quoted by Indian Express, is the denial without the box: no change to crawler behavior. A denial of a behavior change can sit next to a real edit of a contract. The edit is the news. The denial is Meta's account of what the edit means.Futureweb.

A contract edit is not a crawl log

Removing a sentence changes what the terms say. It does not, by itself, show what a crawler fetched last week or will fetch next week. Roberts says behavior does not change. The critics had said the old sentence could be used even if behavior had not yet changed. Both claims can be argued from the text. Neither is a server log. This page will not invent a before-and-after of pages fetched. It will keep the terms edit and the company's statement that robots.txt still applies.the key stays with the holder.

October 30 is a different line

Yesterday's brief was about commercial terms described as taking effect on October 30, and about two readings of the crawl sentence. The October 17 reports say that sentence was removed before that date arrived. If the rest of the October 30 terms still take effect, they take effect without the sentence these outlets quote as deleted. This page will not assume every other clause was withdrawn. The outlets do not say that.

Who the commercial terms bind

Businesses with Meta accounts, which is the population the commercial terms address, including publishers who were the critics' example. A reader with no business account is not the party to that contract. The privacy question for everyone else is unchanged by a clause they were never asked to accept. Asked directly, What did the deleted sentence say, according to The Verge? That Meta could use any content from the domains, pages, or URLs tied to a business to provide, develop, improve, research, train, and maintain Meta products. The Verge says that wording no longer appears on the commercial terms page.the paper.

What this page will re-check

The commercial-terms URL on October 30, for whether the quoted sentence is still absent and whether other clauses moved. A Meta post that publishes the revised paragraph in full, rather than a spokesperson's summary. A publisher who shows a crawler ignoring robots.txt after the edit, which would test Roberts's sentence against a log. That log is not in today's sources. Asked directly, Who asked for it to come down? The outlets credit criticism from Nieman Lab and from James Grimmelmann, a Cornell law professor, who read the sentence as permission to train on scraped pages and to override a robots.txt block. That reading is theirs. The deletion is what the outlets report happened on October 17.an evaluation a second lab can run.

Three outlets, one deletion

The Verge, TechCrunch, and Indian Express are not three independent crawls of Meta's network. They are three newsrooms reporting one contractual edit and one company statement. Agreement among them is strong evidence the sentence came down. It is not a measurement of training data. Treating the deletion as proof that models were retrained, or were not, would add an event nobody reported.

What stays specific

October 17, 2026. The train-on-linked-pages sentence, as The Verge quotes it, reported removed. Roberts: ads and business products, crawlers unchanged, robots.txt still the signal. Stone: no change in crawler use. Yesterday's brief remains the account of the sentence when it was present. The short answer this page will stand on is this: The Verge quotes a commercial-terms sentence, now absent, that said Meta could use content from a business's linked domains to provide, develop, research, and train its products. TechCrunch and Indian Express report the same deletion on October 17. Meta spokesperson Daniel Roberts told The Verge the consolidated crawling terms applied to ads and business products, that crawler behavior did not change, and that robots.txt still signals a preference.