Brazil halts Meta's generative AI training over data privacy concerns

Meta, formerly known as Facebook, previously received a five-day extension from Brazil's data protection agency, ANPD, to comply with an order suspending the use of personal data for training generative AI systems. A Meta spokesperson recently communicated with TechCrunch, saying that Meta will suspend genAI features that were previously live in Brazil while they engage with the ANPD to address their questions around genAI.

The initial deadline, set for July 9th, 2024, was deemed insufficient by Meta, which had requested additional time to implement the necessary measures. The company previously submitted a proposal with additional steps they intend to take to address the ANPD's concerns.

The ANPD's first order in July 2024 mandated Meta to halt the use of personal data for training generative AI systems within Brazil. The agency stated that Meta could leverage this data to train AI systems capable of harming individuals. Naturally, Meta contested the ANPD's decision, arguing that the suspension was unnecessary and detrimental to their business operations. However, the ANPD upheld its ruling.

Data protection is becoming key as artificial intelligence is slowly (but steadily) inserting itself into our lives. Quite recently, Meta's prospective AI partnership with Apple was shot down over privacy concerns. Shortly thereafter, Apple collaborated with OpenAI instead. It's important to note that the ANPD's decision is specific to Meta and doesn't apply to other companies, but this could have wider implications for other companies training their AI systems.