Ethics and the Use of Artificial Intelligence by Lawyers: Limits and Responsibilities

This page is a translation of the original Portuguese text, provided for convenience. It deals with Brazilian law. If there is any discrepancy, the Portuguese version prevails. Read the original in Portuguese

The growing use of automated tools in law firms has boosted productivity, but it has also revived an old debate about the ethical limits of professional practice. Experts in disciplinary law maintain that technology can support a lawyer’s daily work but can never replace the lawyer’s own judgment. The lawyer remains personally responsible for checking the results, safeguarding confidentiality, and any flaws in the work delivered to the client.

Automation does not shift the lawyer’s responsibility

The starting point of the discussion is simple, and it is also decisive. Automated tools process data, organize information, and suggest wording, but they do not take on professional duties. Those duties arise from the relationship between a lawyer and the client who hires them, and they cannot be delegated. When a system produces a flawed draft, the error does not belong to the machine. It belongs to the professional who used it and signed off on the result.

The Statute of the Brazilian Bar (Estatuto da Advocacia), Law 8.906 of 1994, makes lawyers responsible for the acts they perform in the course of their profession. The statute draws no distinction between work done with pen and paper and work supported by technology. The practical consequence is straightforward: using automation does not create a zone of non-liability, nor does it serve as a shield for negligence or inattention toward the client and the courts.

This view fits within the disciplinary logic long established in Brazilian legal practice. The trust a client places in a lawyer rests on a qualified person, not on a piece of software. Any gain in efficiency must therefore coexist with full adherence to the ethical commitments that have structured the profession since its origins. The convenience of new solutions is no excuse for relaxing standards that have always come with the job.

The duty to verify: technology as a means, never an end

Among the obligations that take on greater importance in this setting is the duty to verify. Automatically generated results call for careful review before any use in a court proceeding. Case citations, statutory references, calculations, and deadlines must be checked against official sources. Otherwise, the lawyer risks putting inaccurate or simply nonexistent information into a filing.

The risk is real. Automated systems can produce content that looks precise but has no real basis. A plausible case number, a well-written summary of a ruling, or a statutory provision cited with confidence does not guarantee accuracy. The lawyer must verify every statement, because it is the lawyer’s signature, not the system’s, that gives authenticity and accountability to a document placed in the court record.

Courts in several countries have already sanctioned lawyers who submitted filings containing references fabricated by automated tools. The episode serves as a universal warning: verification is not a side task but the core of legal work. Technology speeds up production, yet validating the content remains a human task. It cannot be transferred to anyone else, and it is subject to disciplinary oversight.

There is also a technical competence dimension. A professional who decides to use a particular tool must understand its limitations, know where the data feeding it comes from, and calibrate how much confidence to place in each result. Using a resource without understanding it is itself a form of negligence, and a claim of ignorance is unlikely to relieve of responsibility someone who chose the solution and benefited from it.

Automated tools process data, organize information, and suggest wording, but they do not take on professional duties.

Professional confidentiality and data protection under automation

The second critical area involves professional confidentiality. The inviolability of information a client entrusts to a lawyer is a pillar of legal practice. It is protected both by the Statute of the Brazilian Bar and by criminal law, which makes breach of professional secrecy an offense under Article 154 of the Penal Code. This duty does not weaken when sensitive data is entered into automated tools.

When feeding a system with filings, documents, or a client’s account of events, the lawyer must assess where that information will be processed and stored, who will have access to it, and how long it will remain on record. Carelessly transferring confidential data to third-party platforms can amount to a breach of confidentiality, with disciplinary, civil, and even criminal consequences.

The General Data Protection Law (Lei Geral de Proteção de Dados), Law 13.709 of 2018, reinforces this need for care. A firm that handles clients’ personal data acts as a “controller” and is responsible for the security of that information, including with respect to the tools it engages. Principles such as purpose, necessity, and security guide the choice of technology and call for caution in selecting vendors and configuring each service.

In practice, it is advisable to anonymize data whenever possible, to prefer solutions that offer contractual guarantees of confidentiality, and to avoid entering identifiable information into systems whose operation is not transparent. Confidentiality, it is worth remembering, protects the client, not the lawyer, so there is little room for improvisation.

Responsibility for errors generated by technology

The third area concerns responsibility for failures. When an automated result leads to an error that harms the client, liability is determined under traditional standards. The question is whether the professional acted with the required diligence, checked the material, and took the precautions expected of someone who has mastered the craft.

A lawyer’s liability is usually analyzed as an “obligation of means,” which requires diligent conduct rather than a guarantee of a specific outcome. Even so, failing to review automated content, missing a deadline through misplaced trust in a system, or inadvertently disclosing confidential data can amount to negligence and give rise to damages, in addition to disciplinary sanctions.

The contractual dimension must be added to this. A client who hires a professional expects personal dedication and care. Quietly handing a significant part of the work to a tool, without adequate review, defeats that legitimate expectation and may constitute a breach of the duties the lawyer assumed, especially when there is concrete harm causally linked to the lawyer’s omission.

The right balance, then, lies neither in rejecting technology nor in embracing it without discernment. It lies in integrating it transparently while keeping the professional at the center of decisions. Used well, automation frees up time for strategic analysis. Used poorly, it becomes a source of ethical and financial risk for the firm itself, jeopardizing a reputation built over years of practice.

Frequently Asked Questions

Can a lawyer use automated tools to draft filings?

Yes. There is no prohibition on using technology to support legal work. What is required is that the lawyer fully review the content, check citations, deadlines, and legal grounds against official sources, and take personal responsibility for the material delivered. The tool assists, but the decision and the validation remain with the lawyer.

Can entering client data into automated systems violate confidentiality?

Yes, it can, when the data is confidential and the platform does not offer adequate safeguards. The duty of confidentiality is grounded in the Statute of the Brazilian Bar and in criminal law, and the protection of personal data is reinforced by Law 13.709 of 2018. You should anonymize sensitive information and consider carefully where and how the data will be processed.

Who is responsible for an error generated by an automated tool?

The lawyer who used the tool and signed off on the result. Responsibility does not transfer to the system, because professional duties are personal and cannot be delegated. Failing to verify the output, missing a deadline through misplaced trust, or disclosing confidential data can lead to civil liability and disciplinary sanctions, depending on the diligence shown in each case.

As informações deste artigo são de caráter informativo e não substituem consulta jurídica individualizada. Cada caso possui particularidades que exigem análise específica.