Meta kept generative AI moving with a regional opt-out policy

In September 2023, Meta said public Instagram and Facebook posts—including photos and text—had been used to train the models behind new generative-AI features. By July 2024, Latin American artists said they had not been notified and could not find the opt-out form offered to users in the EU and U.K.

The turning-point decision was to apply a different data-rights and rollout path by jurisdiction. Brazil’s privacy framework required an opt-out, while several Spanish-speaking countries had laws that were outdated or absent; the article says Bolivia and Paraguay had no general data-protection laws.

That split let Meta continue building and releasing features in the U.S., Australia, Canada, Ghana, South Africa, India and other markets, while its EU rollout was paused after regulator intervention. It preserved deployment momentum, but left many Latin American users without the same control and pushed some illustrators toward Cara.

Meta trained models on public platform content and launched generative AI features across multiple markets while its EU rollout was paused.

Meta already held a large corpus of public photos and text on its platforms.

European data-protection rules gave users a formal right to object.

Different national privacy regimes created room to proceed without the same opt-out process everywhere.

Artists relied on Instagram to reach clients, so abandoning the platform was costly.

When regulation differs by market, a geographic rollout can preserve momentum—but unequal user rights can create legitimacy and creator-trust costs that constrain expansion.

Brazil’s National Data Protection Authority announced a suspension and a possible daily fine of $8,836 if Meta did not comply. In Europe, the AI rollout had been paused; in other Latin American markets, artists remained exposed to the use of their work and some moved to Cara. Meta said public-data training was industry-wide.