BrazilMarco Legal da Inteligência Artificial (PL 2338/2023), direitos de autor e conexos
Projeto de Lei nº 2.338/2023, arts. 62 a 65 (aprovado pelo Senado Federal em 10 de dezembro de 2024)
Before the second chamber, dated , as of .
An AI training data rule.
As of .
Where it has got to
The text described here is Autógrafo do Senado Federal, the text approved by the Senate on and sent to the Chamber of Deputies. That print is PL 2338/2023, published .
Locally, this stage is Comissão Especial da Câmara dos Deputados, aguardando parecer do relator (Special Committee of the Chamber of Deputies, awaiting the rapporteur's report).
The stage above is recorded at www.camara.leg.br.
The Senate approved a substitute text on and the Chamber of Deputies received it on . The Chamber's record shows the bill in its Special Committee awaiting the rapporteur's report, with other artificial intelligence bills joined to it through September 2026, and no committee report or plenary vote recorded.
What it requires
- This bill has not been enacted and binds no one yet; as the Senate approved it, a developer that uses copyright-protected content in developing an AI system, including its mining and training stages, would have to publish a summary of that content on an easily accessible website, subject to trade and industrial secrecy and a specific regulation.
- If the bill is enacted as approved, stop using a rightsholder's content in developing an AI system once the holder has prohibited that use, unless the non-commercial research, education and cultural-institution text and data mining exception of Article 63 applies.
- If the bill is enacted as approved, remunerate the holders of protected content you use in mining, training or developing an AI system; the holders may negotiate the use collectively or directly, and the remuneration is due to holders domiciled in Brazil and to holders from countries that give reciprocal protection.
These lines are LexLint's reading of the law's text, as of .
What this law does
Article 62 of the Senate-approved text requires an AI developer that uses content protected by copyright and related rights to report on the protected content used in developing AI systems by publishing a summary on an easily accessible website, subject to trade and industrial secrecy. Its sole paragraph states that, for this chapter, development covers the mining, training, retraining, testing, validation and application stages of AI systems.
Article 63 provides that automated use of protected content in text and data mining for AI research and development is no offense to copyright when done by scientific, research and educational organizations, museums, public archives and libraries on conditions of lawful access, no commercial purpose and use limited to what the objective needs. Article 64 lets a holder of copyright or related rights prohibit the use of their content in developing AI systems in the cases Article 63 does not cover.
Article 65 requires an AI agent that uses protected content in mining, training or development of AI systems to remunerate the holders of that content. Article 65, paragraph 1 makes the remuneration due only to rightsholders domiciled in Brazil and to persons domiciled in a country that assures equivalent reciprocal protection.
When LexLint raises it
When your app profile says your app trains models.