On November 29, 2024, the Brazilian data protection authority (ANPD) published a report on generative artificial intelligence (AI) as part of the ANPD's technological radar.

What are the main aspects of the report?

The ANPD highlighted that generative AI models are different from other models as they aim to generate synthetic data with the same statistical characteristics as real data. The ANPD further noted the two different approaches to generative AI and how they operate: generative adversarial network (GAN) and generative pre-trained transformers (GPT).

In relation to data protection, the ANPD noted that generative AI models need large volumes of data, which directly affects the processing of personal data and the principles under the General Data Protection Law (LGPD). The ANPD stated the possibility of generative AI systems processing both personal and non-personal data, which increases the likelihood that personal data will be processed without the appropriate safeguards and that the principle of necessity will not be met.

In light of that, the ANPD highlighted some measures to help ensure that personal data is protected in generative AI systems, such as anonymization, transparency, and the establishment of a chain of responsibility for all entities involved in the collection and processing of personal data.

The ANPD also noted some use cases of generative AI in the national context.

You can read the press release here and the report here, both only available in Portuguese.