By Carolina Pacheco de Andrade
Despite the recent popularization of artificial intelligence (AI) tools and some initial resistance to these technologies, it is now difficult to argue that we perform tasks, even the simplest ones in daily life, without any influence, information, or assistance from these tools.
What the numbers show
The AI Diffusion Report, published by Microsoft, indicates that the use of generative AI grew in the first quarter of 2026, rising from 16.3% to 17.8% of the world’s working-age population.
Brazil has stood out in this scenario. According to the AI Sentiment Index, a global ranking created by Ernst & Young to measure the use of artificial intelligence, 95% of respondents in the Brazilian sample say they already use AI, ten percentage points above the global average, and 21% say they use AI agents, which operate autonomously without the need for human involvement.
The legal field is no different. AI is already applied in the day-to-day work routine, including through tools specialized in the sector. The 2026 edition of the Report on the Impact of AI on the Law, developed by the São Paulo, Paraná, Bahia, Goiás, Pernambuco, and Espírito Santo chapters of the Brazilian Bar Association (OAB), in partnership with Trybe, Jusbrasil, and ITS Rio, indicates that 77% of the legal professionals consulted say they frequently use AI tools at work, up from 55% in the previous edition. At the report’s launch, the Vice President of the São Paulo Bar Association (OAB/SP), Daniela Magalhães, emphasized:
“We are turning this transformation into a form of responsible support and an ethical reflection on the use of artificial intelligence.”
On the duty of technical supervision
This growing adoption coexists with a structural limitation of the technology: AI tools do not guarantee the accuracy of the information they produce. In the legal field, it is not uncommon to find, in work generated by these platforms, references to statutes, decisions, or interpretations that do not exist, a phenomenon known as hallucination. This risk is not incidental, but inherent to how these tools operate: by statistical language probability, not by source verification.
It follows that legal work conducted through AI without technical supervision may contain errors that, even if they go unnoticed at first, can cause significant harm. Review by a qualified professional is what averts this risk: it is the lawyer’s responsibility to analyze and confirm the information provided by the tool and to bear technical responsibility for the outcome from start to finish. This is not merely a recommendation of prudence, but a professional duty of competence and diligence that applies regardless of the means used to produce the work. In this regard, Recommendation No. 001/2024 of the Federal Council of the Brazilian Bar Association (OAB), in item 3.3, provides that:
“Excessive reliance on AI tools is inconsistent with the practice of law and cannot replace the analysis performed by the lawyer.”
On confidentiality and data protection
The use of AI also requires reconciling it with professional secrecy and data protection rules. Entering client information into third-party platforms in practice means sharing that data with the tool’s provider, which calls for attention to two overlapping regulatory frameworks: the lawyer’s duty of confidentiality, set out in Article 7, II, of Law No. 8,906/1994 (the Brazilian Bar Statute), and the regime of the Brazilian General Data Protection Law (Law No. 13,709/2018, the LGPD).
Recommendation No. 001/2024 itself, in item 2, devoted to confidentiality and privacy, addresses the matter as follows:
2.1. When entering information into AI systems, the lawyer must safeguard the confidentiality and professional secrecy of the data submitted, paying particular attention when entering data that could inadvertently make the client identifiable.
2.2. Diligence is required in selecting the AI system, to ensure that the product’s provider will protect the information entered into the system, adopt security measures, and allow the data provided not to be used to train the systems.
2.3. The lawyer must remain alert to data sharing by AI systems, making sure that the purposes of such data sharing are duly set out in the privacy policy.
2.4. The use of virtual assistants (chatbots) must not include the performance of activities reserved exclusively to lawyers, and the other party must be clearly informed that they are interacting with a machine.
2.5. Protecting client information is essential. Law No. 13,709/2018 must always be observed before entering confidential data into AI tools.
Along the same lines, item 4.1.1 provides that a lawyer who chooses to use AI tools in rendering services must formalize that intention to the client before starting to use them. Transparency, here, is not a courtesy but an extension of the duty to inform that governs the lawyer-client relationship.
Support, not substitution
Taken together, these guidelines reinforce a single understanding: AI should not be seen as a threat, but as a layer of support for professional practice. Technology shifts the value of legal work away from mechanical execution and toward critical judgment. Tasks such as research, screening, and organization, once time-intensive, tend to be accelerated; in turn, analysis, contextualization, and technical responsibility, which remain non-delegable, gain greater weight.
Thus, the differentiator no longer lies in access to the tool, now widely available, but in the ability to use it with judgment: knowing what to ask, recognizing the weaknesses in the response, and standing technically behind the result. It is in this balance between technological efficiency and human critical analysis that the practice of law defines itself in the face of AI.
GTLawyers follows this movement and incorporates it into its practice, always subordinating the technology to the review and technical responsibility of its lawyers.






