On 3 November, a court in São Paulo state, Brazil, issued a preliminary injunction against Meta's new Commercial Terms: news content from members of three national press associations may not be used to train AI without authorisation. On 6 November, Meta announced it was pausing the update, and the planned 20 November effective date is now void. It is the first time a Brazilian court has blocked Meta's scraping terms with an injunction, with fines of R$1 million per day (about US$180,000) for non-compliance, capped at R$100 million.
What the injunction blocks
On 3 November, Judge Guilherme Madeira Dezem of the 44th Civil Court of São Paulo granted the injunction at the request of Brazil's National Newspaper Association, the Radio and Television Broadcasters Association and the Magazine Publishers Association. The order suspends the relevant parts of Meta's new Commercial Terms and bars Meta from collecting or using member outlets' content without authorisation, including using it to develop, train or feed Meta AI systems; it also bars Meta from suspending accounts, restricting pages or cutting off business tools because an outlet refused the new terms. The judge wrote that requiring media to hand over copyrighted work for free in exchange for staying on the channels through which the public now gets its news is a manifest abuse of rights. The ruling also notes that Brazilian law has no general text-and-data-mining exception for commercial AI training, so reproducing and commercially exploiting a work requires the rights holder's authorisation. Meta must respond within five days of being notified.
Who is affected, and what to do now
The injunction directly protects only the member outlets of the three associations, and only in Brazil — it is not a worldwide ban. Meta's 6 November pause covers the terms update itself, spanning four sets of terms, and the company says it will keep engaging with stakeholders while the existing terms remain in effect. Member outlets can refuse the new terms without fear of their accounts being penalised; outlets outside the associations have no protection from the injunction and must assess the risk themselves. Businesses should note that Meta's preview of the terms no longer uses the word "train", but the crawling permission remains: businesses authorise Meta to use automated software such as scrapers to "access, index, cache, analyze, collect, retrieve, or crawl" the domains linked to their products or services. Companies using Meta's business products to drive traffic should check whether their own sites fall under that permission, keep records of authorisations and communications, and run a separate compliance review for their Brazil operations.
Risks and limits: a pause is not an end
The injunction is only a provisional measure, not a final judgment; the outcome may change once Meta presents its defence, and no date has been set for a final ruling. Similar pushback has already happened: on 28 October, the Oslo District Court temporarily froze the same terms at the request of ten media groups; the UK Competition and Markets Authority has opened an investigation into Meta over the same terms; and Meta had already removed the word "train" from the terms and pushed the effective date to 20 November. The 6 November pause makes the 20 November date moot for now, but the dispute over the crawling permission itself is unresolved — the update is only paused, and Meta has not withdrawn the crawling framework. Two things to watch next: how Meta responds to the Brazilian court within its five-day window, and whether its talks with publishers produce a new draft of the terms.
