Yesterday, December 10th, 2024, the Brazilian Senate approved Bill No. 2,338/23, which aims to regulate artificial intelligence (AI) systems in Brazil.
Among the main points of the Bill, we highlight:
- Creation of the SIA – National System for AI Regulation and Governance;
- Establishment of rights for individuals or groups affected by AI;
- Creation of AI governance rules;
- Definition of excessively risky AI, which development is prohibited;
- Definition of high-risk AI, subject to stricter compliance rules;
- Establishment of different civil liability regimes;
- Imposition of fines of up to R$ 50 million or 2% of turnover in case of violations.
Regarding copyright, the Bill limits the “fair use” of copyrighted works for AI training purposes to scientific and nonprofit research organizations. In practice, this means that copyright infringement will occur if AI training is conducted by a private company or a research institution linked to a for-profit organization.
Under the Bill, authors whose rights are violated will be entitled to “reasonable” “proportional” compensation that takes into account the “size of the AI agent”. Moreover, authors may also request a preliminary injunction to immediately stop the infringement. The next step will be the review of the Bill by the House of Representatives. If no amendments are made, the document will proceed to presidential approval. If amended, the text will return to the Senate for a new vote.
Our team of experts is available for further clarification.
