Tech roundup: Experts say AI regulation bill is born dated

Welcome to our Tech Roundup, where we bring you Brazil’s biggest stories in technology and innovation. This week: The buzz around Senate President Rodrigo Pacheco’s bill that promises to regulate the use of AI in Brazil.

For experts, AI regulation bill is born dated

Senate President Rodrigo Pacheco presented in early May a proposal to create a civil rights framework for artificial intelligence (AI), aimed at regulating AI systems in the country. Although drafted by several experts after a series of public hearings and seminars, the bill predates the boom of ChatGPT and other language modeling (LLM) solutions, and may be ineffective in addressing the challenges posed by these technologies.

State of play. The bill requires developers and companies to evaluate and classify their AI solutions according to risk levels, it establishes guarantees of rights for people affected by these tools, and it defines parameters of responsibility and punishment for the agents involved.

• The proposal also creates a government agency linked to the president’s office to provide corporate oversight — similar to the authority designed to lead the country’s data protection law. This agency would be responsible for deciding, for instance, when audits are needed to assess the impact of a given AI platform.
• Under the bill, companies would be required to inform individuals of the existence of emotion recognition or biometric categorization systems, and explain how they work.

Punishment. Companies that fail to comply with the law could face various types of sanctions, including the interruption of certain innovation projects for up to five years or the indefinite suspension of all activities. Fines are capped at BRL 50 million (USD 10.1 million) or the equivalent of up to 2 percent of the company’s revenues.

How the bill was drafted. Some critics...