Brazilian Court Restricts Meta’s Use of Journalistic Content for AI Training

A recent preliminary injunction issued by a court in São Paulo suspended provisions in Meta’s new commercial terms that would allow journalistic content to be collected and used for the development and training of artificial intelligence systems.

The lawsuit was filed by Brazil’s main associations representing newspapers, broadcasters and magazine publishers. The challenged terms reportedly required media companies to allow automated tools to access, copy, store and analyze their content, including material hosted outside Meta’s platforms, as a condition for continuing to use those platforms.

The authorization was broad, covering news articles, photographs, illustrations and audiovisual works, including previously published content, for the development and improvement of commercial products and AI systems. It could also affect third-party content licensed to media companies for specific purposes.

As an interim measure, the court suspended the challenged provision and prohibited the unauthorized collection and use of content belonging to the member companies. It also rejected the application of the foreign forum selection clause and confirmed the jurisdiction of Brazilian courts.

Meta was further ordered to keep the companies’ accounts and tools on Facebook, Instagram, WhatsApp and Threads operational, without restrictions resulting from their refusal to accept the new terms. A daily fine of BRL 1 million was imposed for non-compliance, capped at BRL 100 million.

Although the decision may still be reviewed and applies within the specific context of this lawsuit, it raises a broader issue: when can copyright-protected content be collected and used to develop AI tools?

The case suggests that simply including a broad authorization in terms of use may not be enough, particularly where the authorization:

• does not clearly identify the content and permitted uses;

• extends to material hosted outside the relevant platform;

• includes third-party rights or content;

• is required as a condition for access to commercially important platforms; or

• allows content to be used for purposes not originally contemplated by the rights holder.

The decision also highlights an important distinction between having technical access to online content and having the legal right to reproduce, store, transform or use it. The fact that content is publicly available does not, by itself, mean that it is free to use for AI training.

What does this mean for international businesses?

The issue is not limited to the media industry. Companies in many sectors create, license and publish content that may be used to train or improve AI tools, including photographs, videos, technical materials, catalogs, advertising, databases, manuals, designs, illustrations and software.

For companies operating in Brazil, this makes it increasingly important to understand what content may be accessed by AI tools, whether the necessary rights and licenses are in place, and what technical or legal measures may be available to control its use.

For international counsel advising clients on cross-border AI, copyright and technology matters, the Brazilian aspects should also be considered separately, particularly where content is created, licensed, published or collected in Brazil.

Our team is available to work alongside foreign counsel on the Brazilian aspects of AI and copyright matters, including the review of contracts and terms of use, assessment of rights in protected content, and strategies for the protection and licensing of these assets under Brazilian law.