BRAZIL Trends and Developments Contributed by: Eduardo Perazza, Glaucia Coelho, Débora Chaves Martines Fernandes and Agnes Domingues, Machado Meyer
only slightly for legal research (59%), the drafting of opinions and memoranda (58%), and contract analy- sis and review (56%), suggesting that AI has penetrat- ed virtually every dimension of legal practice. Perhaps most notably, it is not merely that lawyers are working faster, but they believe they are working better. 91% of professionals who use AI report improvements in the technical quality of their work. Whether that per- ception reflects a genuine improvement in quality or a more subtle effect of fluency and confidence that AI-generated text tends to project is a question the data alone cannot answer. What the data confirms is that AI has moved well beyond peripheral experimentation in Brazil. This is a shift consistent with a broader global phenomenon: according to the American Bar Association’s Legal Technology Survey Report 2024, AI adoption by US law firms nearly tripled in a single year, with signifi- cant growth among solo practitioners and small firms, indicating that the technology has ceased to be the prerogative of large firms and has become part of the routine of professionals of the most varied profiles and structures. Accessible tools – many of them free or low-cost – have made it possible for any lawyer, regardless of the size of their firm, to incorporate the technology into their daily work. The emerging standard: technological competence as professional duty The rapid adoption of AI on both sides of the court- room – by judges drafting decisions and lawyers pre- paring filings – creates a new and largely unchartered dimension of professional responsibility. When 76% of lawyers use AI to draft procedural documents and courts employ generative AI to screen and draft rul- ings, the question is no longer whether practitioners should engage with the technology but whether fail- ure to do so competently constitutes a professional shortcoming. This shift echoes a debate that has already matured in other jurisdictions. In the United States, over 80% of state bars have amended their rules of professional conduct to include a duty of technological compe- tence, requiring lawyers to remain informed about the benefits and risks of technology relevant to their prac- tice. In Brazil, no equivalent formal obligation exists
yet. The OAB-SP Report, however, reveals that over 60% of lawyers who do not use AI cite lack of knowl- edge as the primary barrier, not cost nor scepticism. The gap, in other words, is not one of willingness but of training. For clients with high-stakes disputes before Brazil’s Superior Courts, this emerging competence divide carries tangible consequences. A lawyer who under- stands the environment in which they litigate – includ- ing how automated screening operates – is better positioned to craft filings that are not only substan- tively sound but also structurally visible to the system that will first process them. A lawyer who ignores this dimension of the proceedings may produce argu- ments that are legally correct but that fail to engage the criteria the automated filter is designed to detect. Whether practitioners should be expected to opti- mise their submissions for the machine, rather than solely for the law, remains an open and uncomfort- able question, but as automated admissibility screen- ing becomes the effective gateway to the STJ, it is a question that the profession cannot afford to leave unanswered. Hallucinations, sanctions and a regulatory framework still taking shape The adoption of AI by courts and lawyers in Brazil has moved far faster than regulation and faster than col- lective understanding of what the technology does. As in the rest of the world, users interact with genera- tive AI through a clean interface that masks extraor- dinary complexity underneath. The output is fluent, well-structured and apparently authoritative – it cites cases, invokes statutes, builds arguments with logical coherence. That very fluency is what creates a sense of reliability that has not been entirely earned yet. The regulatory landscape Regulation has struggled to keep pace with adoption, but a framework is gradually emerging. In the absence of comprehensive legislation, the CNJ has taken the lead in setting ground rules. Its most significant inter- vention came in March 2025 with Resolution No 615, which replaced an earlier and more general framework (Resolution No 332/2020) and addressed generative AI specifically. The core message of the new resolu- tion is clear: AI may assist but must not replace the
75 CHAMBERS.COM
Powered by FlippingBook