The case, explained

CNF Opinion: Ethical Responsibility in the Use of Generative AI

6 min read · Updated September 2026 · Editorial oversight: Avv. Federico Papa

The landscape of the legal profession has undergone a decisive shift with the consolidation of interpretive guidelines regarding the use of generative artificial intelligence. According to reports from the specialized press in 2024, the National Bar Council (CNF) has clarified that the adoption of technological tools does not exempt lawyers from their fundamental duties; rather, it increases their liability in the event of gross errors. While previous articles concerning the CNF focused on lead generation and the reform of the legal system, this contribution addresses the specific ethical and civil liability deriving from the use of generative AI, with particular regard to legal hallucinations and the duty of human oversight. Opinion no. 18/2024 confirms the need for rigorous control against blind delegation to algorithms. We illustrate this complex subject through a twin case featuring our recurring characters, to make the risks of a powerful but flawed technology tangible.

CNF Opinion: Ethical Responsibility in the Use of Generative AI

In brief

The article examines CNF opinion no. 18/2024 on the ethical responsibility of lawyers using generative AI. It analyzes the duty of professional diligence and the non-delegability of legal research. Through the twin case of Gaio Sventura, it illustrates the civil and disciplinary consequences of legal hallucinations, emphasizing the importance of human oversight, as also required by the European AI Act, to prevent judicial errors and damage to professional dignity.

  1. The facts

    The matter stems from advisory opinion no. 18 issued by the National Bar Council in July 2024, during a period of rapid AI expansion in the legal sector. According to reports by Il Sole 24 Ore and Altalex, the CNF intervened to provide that lawyers are solely responsible for documents bearing their signature, even if generated by software. This is an institutional directive acting as an interpretive guide for District Disciplinary Councils. The press linked this stance to the well-known US case where a lawyer produced briefs containing case citations entirely invented by the algorithm. In Italy, the reflection went further, qualifying legal research as an intellectual activity that cannot be outsourced to machines without critical and constant control, under penalty of breaching the fiduciary relationship with the client.

  2. The laws in play

    The regulatory framework revolves around the Code of Legal Ethics and the Civil Code.

    1. Art. 12 of the Ethics Code imposes the duty of diligence, which in the case of AI translates into the obligation to verify every single citation produced by the software.
    2. Art. 1176, paragraph 2, of the Civil Code establishes professional diligence higher than that of the average person, making the lawyer liable for avoidable technical errors.
    3. Art. 14 of the Ethics Code on competence forbids accepting assignments without having the technical tools to manage them, including knowledge of AI limits.
    4. The AI Act (EU Regulation 2024/1689) introduces the principle of human oversight to mitigate automation risks in critical decision-making processes.
  3. What case law says

    Although there are no definitive Supreme Court rulings specifically on algorithmic hallucinations, established case law has long clarified that lawyers are liable for the work of their assistants. According to this consolidated trend, a professional using a collaborator or a technical tool is bound to a control and review activity that absorbs the tool's error. Lower court jurisprudence has also specified that citing repealed or non-existent laws constitutes gross negligence, excluding the limitation of liability provided for cases of special technical difficulty, as verifying sources is a basic activity of the profession. The principle of self-responsibility of the person signing the document remains the fundamental pillar of legal defense.

  4. Analysis drafted and verified with edit.legal

    To verify the provisions cited in this article, we used edit.legal. Test our legal AI on official sources and apply it to your own matters.

    Try edit.legal AI
  5. What it teaches professionals

    1. Never trust algorithmic output without cross-verification on official legal databases.
    2. Include a transparency clause in professional mandates regarding the use of AI tools to ensure the client's informed consent.
    3. Constantly train on the logic of language models to promptly recognize error patterns.
    4. Adopt internal firm policies for the supervision of collaborators and software used in drafting legal documents.

References: Art. 9 Codice Deontologico ForenseArt. 12 Codice Deontologico ForenseArt. 1176 Codice CivileArt. 2236 Codice CivileRegolamento UE 2024/1689 (AI Act)

Avv. Federico Papa
Editorial oversight: Avv. Federico Papa·ICAMContent drafted with AI support and subject to editorial source checks. Despite these controls, inaccuracies may remain: reports and rectification requests are welcome. Report a correction

Frequently asked questions

What happens if a lawyer cites a non-existent case due to an AI error?

The lawyer risks a disciplinary sanction for breach of the duty of diligence and can be held civilly liable for damages caused to the client due to gross negligence.

Must the use of AI be communicated to the client?

The CNF opinion suggests transparency and fairness; it is advisable to inform the client about the use of such tools within the professional mandate disclosure.

Can AI replace traditional legal research?

No, AI can only be an initial support; definitive research and source validation must be performed personally by the professional to ensure data accuracy.

Verified legal research and drafting with edit.legal

Legal research and drafting with citations checked against official databases. edit.legal is free to try, no credit card.

Try edit.legal for free