The debate on introducing Artificial Intelligence (AI) into the justice system and the legal profession has moved beyond mere theoretical speculation and forcefully entered the courtroom.
While, on the one hand, Large Language Models (LLMs) and legal tech software promise an unprecedented revolution in terms of time optimisation, document analysis and case-law research, on the other, the risks of uncritical and unregulated reliance on these tools are becoming starkly apparent.
The judiciary and lawmakers, at both national and European level, are drawing very clear boundaries: Artificial Intelligence can and must be an aid to legal professionals, but it can never replace human critical reasoning, empathy and professional ethical responsibility.
The judiciary’s censure: the “vexatious litigation” case
Italian judges have already strongly censured the misuse of AI in drafting defence submissions.
The jurisprudential turning point is a well-known and landmark order of the Court of Turin, issued in 2025. In that case, the judge had to examine a claim that had plainly been generated by generative Artificial Intelligence software.
At first glance the filing appeared structurally flawless, but on closer examination it proved to lack any genuine line of argument, was self-contradictory and, above all, was riddled with entirely fabricated statutory and case-law citations (the well-known phenomenon of algorithmic “hallucinations”).
The Court did not merely dismiss the claim on the merits; it also sanctioned the party and its counsel for aggravated liability under Art. 96 of the Code of Civil Procedure (vexatious litigation).
The reasoning of the Piedmont judges established a cornerstone principle: the use of AI does not release counsel from the duty to verify sources.
Filing a document containing false precedents or reasoning devoid of legal logic distorts the function of the defence and undermines the principle of a fair trial, as well as placing a needless burden on the judicial system.
The judges’ censure highlights how technological copy-and-paste breaches the duty of professional diligence and the principles of fairness and probity in proceedings.
The Italian and European regulatory framework: Bill 1146 and the AI Act
To prevent innovation from turning into a legal Wild West, the institutions have set out a rigorous regulatory framework.
At European level, the AI Act (Regulation (EU) 2024/1689) has classified Artificial Intelligence systems intended for use by judicial authorities or in the administration of justice as “high-risk” systems. This means such software must meet extremely stringent requirements on transparency, traceability, cybersecurity and, above all, constant human oversight.
In Italy, lawmakers have taken up these demands through the Artificial Intelligence bill (Ddl 1146/2024, consolidated into Law No. 132 of 2025).
The legislation devotes a specific section to the use of AI in the administration of justice and in the intellectual professions.
The law expressly states that AI systems may be used solely for organising office work, document research and interpretative support.
However, it sets an absolute prohibition: AI may never replace the judge in assessing facts and evidence, nor in deciding the dispute.
Likewise, for lawyers, the legislation establishes that the contribution of technology must remain merely instrumental and complementary to the professional’s intellectual work. The lawyer remains solely and primarily responsible for the procedural document, towards both the client and the court.
The law also imposes a principle of transparency: significant use of generative AI tools in drafting documents or providing advice must be disclosed, ensuring the traceability of the algorithm used.
The irreplaceable value of lawyers and judges
Case law and legislation converge on one essential point: a judicial decision and a defence strategy are not the product of mere probabilistic calculation or statistical correlation of past data.
Law is a living social science, evolving as a society’s civil, moral and economic sensibilities change.
An algorithm is, by its nature, “retrospective”: it analyses the past to predict or replicate behaviour in the present.
Had we relied on predictive Artificial Intelligence over recent decades, landmark developments in civil rights, family law or civil liability would never have come about, because the algorithm would simply have replicated the majority (and outdated) position of the past.
Moreover, criminal and civil proceedings require human empathy.
Understanding the context of an offence, assessing a witness’s credibility not only from their words but from their non-verbal language, or balancing the interests at stake in a complex international corporate dispute or a family crisis, requires a sensitivity that eludes any string of code.
Conclusions: towards a conscious “Legal Engineering”
The outlook should not be read as a wholesale rejection of technology and innovation.
Artificial Intelligence is an extraordinary tool when integrated into a modern “legal engineering”. E-discovery software, document classification systems and virtual assistants for standardised contracts can free judges and lawyers from repetitive, low-value tasks, allowing them to focus on the true core of the profession: developing strategy, interpreting the law creatively and conducting courtroom argument.
The real challenge for 2026 and the years ahead lies in training.
Universities and Bar Councils must train lawyers with solid digital skills, able to interact with machines and to subject algorithmic outputs to rigorous critical scrutiny.
Only by keeping human beings at the centre of the system, as guardians of legality, ethics and accountability, can Artificial Intelligence turn from a threat to legal certainty into a powerful engine of efficiency and justice.
About Marco Tupponi
A lawyer admitted to practise before the Italian Supreme Court, specialising in national and international commercial law, Technology Law and artificial intelligence, he is an adjunct professor at the University of Bologna, Forlì campus. Founder of Studio Tupponi, De Marinis, Russo & Partners, he works with ICE, the Ministry of Foreign Affairs, the Ministry of Enterprises and Made in Italy and Il Sole 24 Ore. He is also a member of the ICC Commission on Tech Law and AI and the author of numerous publications, including the Manuale di Diritto Commerciale Internazionale.
On Bliss Faculty
Bliss Faculty is Bliss Agency’s editorial column featuring signed contributions from leading Italian university lecturers and researchers. Each article draws on its authors’ scientific expertise and is edited, distributed and optimised entirely by the Bliss team, bringing Italian academic excellence to an online audience.
New Connections (FAQ)
Can a lawyer use ChatGPT to draft a court document?
A lawyer may use ChatGPT or other artificial intelligence systems as support tools to organise information, summarise documents, develop a first draft or carry out research. The professional’s intellectual work must, however, remain predominant, and all content must be verified before signing and filing.
Law No. 132 of 2025 also requires that the client receive clear, simple and comprehensive information on the artificial intelligence systems used in carrying out the mandate. This obligation concerns the relationship between professional and client and does not automatically mean that the use of AI must be stated in every procedural document.
Who is liable if artificial intelligence cites non-existent rulings?
Responsibility remains with the professional who signs and files the document. Using a generative system does not remove the duty to check that sources exist, that precedents are relevant and that the legal reasoning is coherent. Fabricated citations or manifestly irrelevant arguments may lead to the document being rejected, to disciplinary consequences and, where the conditions set by law are met, to an order for aggravated liability under Article 96 of the Code of Civil Procedure. The sanction for vexatious litigation does not follow automatically from the use of AI, but from the negligent or abusive way in which the content is produced, verified and used.
How can a law firm introduce artificial intelligence safely?
A law firm should start from rules that distinguish authorised tools, the data that may be entered, the tasks that can be delegated and those that always require professional review. It also needs procedures to check sources and citations, inform clients, manage access, document the most significant uses and report any incidents. Translating these principles into an AI Governance system for the firm, building an AI Policy, Human Oversight protocols, supplier selection criteria and training programmes consistent with professional practice, is therefore essential.
BLISS®
© 2026. ALL RIGHTS RESERVED