1. News
  2. Everyday tech
  3. Meta · AI and news
Meta · AI and news1 min read

Brazilian court blocks Meta from using news to train its AI

A São Paulo court barred Meta from scraping Brazilian news to train AI, a sign courts may make AI firms ask permission before using publishers' work.

By the Dashline deskSource: Folha de S.Paulo
Brazilian court blocks Meta from using news to train its AI
Photo: Blogtrepreneur, CC BY 2.0

A São Paulo court on Tuesday barred Meta, the owner of Facebook, Instagram, WhatsApp and Threads, from using bots to collect news content from certain Brazilian outlets. It also blocked Meta from using that material to train AI systems such as Meta AI. A Norwegian court suspended the same rules nearly a week earlier.

What the court said

The order covers articles, photos, videos and other content from outlets tied to three Brazilian press associations. The judge said copying news sites to feed AI language models is reproduction of copyrighted works. That needs prior permission, he said, and Brazilian law has no exception for data mining for commercial purposes.

The case targets new Meta terms due to take effect on November 20. The judge suspended the part requiring companies to allow "web scrapers" (automated tools that copy web pages) without payment. He called it contractual abuse. Marcelo Rech, of the Brazilian Newspaper Association, called the ruling "historic."

What happens next

Meta faces a fine of R$1 million ($182,400) a day, capped at R$100 million ($18 million). It has five days to challenge the ruling. The associations have 30 business days to file their main claim. Meta was not heard in court and said it would not comment. It is not yet known whether Meta will challenge the order.

The takeaway

If you publish online, courts may start requiring AI companies to get permission before using your content, which could protect your rights and income.

More articles

You should really read these.

Next step

Let’s talk