AI in Medicine Gets Ethical Rules from Brazil's CFM

CFM Resolution 2.454/2026 sets ethical rules for AI in medicine. Clinical decisions stay human, and civil liability remains governed by statute, not by the Council's rule.

October 2026 2 min read Celso Freitas
Inteligência artificial na medicina ganha regras éticas do CFM

Artificial intelligence in medicine is a support tool, and the clinical decision remains with the physician.

CFM Resolution No. 2,454/2026 (CFM, the Brazilian Federal Council of Medicine) governs, from an ethical and professional standpoint, the use of artificial intelligence in medicine. In practice, the technology serves as support for the physician, and the clinical decision remains a human act.

Does artificial intelligence in medicine change who decides?

It does not. The Resolution treats the system as a tool, not as a substitute for the professional. Accordingly, the physician remains responsible for assessing the result presented by the machine. In other words, the use of AI in healthcare does not transfer the decision to the system.

The rule also addresses matters such as human oversight, risk management, transparency and data governance. Hospitals and clinics already using diagnostic support software should examine these points carefully.

Is the physician released from liability if the system fails?

The Resolution provides protection for physicians against undue liability for failures attributable exclusively to the system. However, this requires evidence of diligent, critical and ethical use of the tool. The rule has limited scope, because a professional resolution does not amend the law.

The civil liability of physicians, hospitals and clinics arises from the Civil Code and from other statutes. Thus, the protection provided within the scope of the Council does not remove the duty to indemnify where the law imposes it. It is our understanding that recording how the tool was used is likely to carry weight in any eventual defence.

Health data remains governed by the General Data Protection Law (LGPD, the Brazilian data protection statute). In other words, the Resolution does not create a parallel regime for such data.

How does the Resolution compare with the AI bill?

Bill No. 2,338/2023 is still under consideration in the National Congress and seeks to establish a general regime for artificial intelligence across various sectors. The Resolution, by contrast, applies to medicine and is not a general framework. The two texts converge on human oversight, risks, transparency and data.

Compliance deadlines and disciplinary penalties are not set out in the material reviewed and depend on a reading of the official text. Those using these tools should check it before adjusting internal protocols and contracts with suppliers.

Read other legal news commented on by the firm and the articles by Moraes de Freitas Advogados.