blog
AI USE IN THE DELIVERY OF MEDICAL SERVICES
29 de April de 2026While a comprehensive Bill to regulate artificial intelligence (AI) in Brazil remains under discussion in the Congress, regulatory initiatives have developed on a sector‑specific basis. In line with prior sectoral initiatives adopted by the National Council of Justice (Conselho Nacional de Justiça – CNJ) with respect to the use of AI in Judicial proceedings, and by the Superior Electoral Court (Tribunal Superior Eleitoral – TSE) regarding the use of AI in electoral campaigns, the Federal Council of Medicine (Conselho Federal de Medicina – CFM) has moved forward with the establishment of a regulatory framework governing the adoption and use of AI in the delivery of medical services, enacting on February 27th, Resolution No. 2,454/2026. This Resolution shall come into force 180 days after its publication.
The Resolution authorizes the use of AI models/systems/applications as support tool in diagnosis, health care treatment, clinical trials, and health care administration, provided that applicable ethical and legal requirements are observed. The use of AI‑based tools should only occur as a support tool and does not diminish or replace the physician’s independent medical judgment with respect to diagnosis or treatment decisions. AI systems are prohibited from directly communicating diagnostic conclusions or treatment recommendations to patients.
Physicians are required to document in the patient’s medical record the use of AI tools in connection with diagnosis or prescribing decisions and must disclose such use to the patient. Patients have the right to object to the use of AI as support to the physician. The regulation reinforces that responsibility for all clinical decisions remains with the physician, who must critically assess and validate any recommendations generated by AI systems.
The Resolution also establishes guidelines regarding the auditability of AI systems, including requirements related to demonstrated effectiveness and risk classification, with scientific performance indicators to be published as reference benchmarks. In addition, developers are required to implement mechanisms to ensure safety, reliability, privacy, and the protection of patients’ personal data, pursuant to a privacy‑ and security‑by‑design approach.
Our law firm is available to provide any further clarification.
