AI Law

This article examines how artificial intelligence reshapes decision-making within the relationship of care and trust between physician and patient. Starting from the four traditional principles of bioethics – beneficence, non-maleficence, autonomy, and justice – it argues that explainability should operate as a fifth principle, articulated through accountability and intelligibility. The analysis considers the risks generated by opacity, cognitive and discriminatory bias, automation bias, and the possible displacement of both medical professionalism and patient self-determination. It then explores the role of meaningful human oversight, transparency, value-sensitive design, and reason-giving in ensuring that AI remains a support rather than a substitute for clinical judgment. The concluding argument is that legally compliant and human-centred explainability is essential both to professional responsibility and to the patient’s free and informed participation in shared care planning.

This contribution offers a reflection prompted by the White Paper edited by ANVUR (the Italian National Agency for the Evaluation of Universities and Research Institutes), in cooperation with ACN (the Italian National Cybersecurity Agency), entitled «Stato della ricerca sulla cybersicurezza in Italia» (The State of Cybersecurity Research in Italy). The White Paper constitutes the first survey of national research in this field, conducted on the basis of the research outputs submitted for the 2020-2024 Research Quality Assessment exercise. Starting from the classification of cybersecurity as a public function, the contribution examines the methodology and findings of the mapping exercise, which identified 1,154 scientific publications relating to cybersecurity (slightly less than 1% of Italy’s overall scientific output) most of them belonging to the disciplinary fields most closely connected with the subject. The article highlights the strengths of the system, including the geographical spread of research activities and the vitality of doctoral education, as well as its structural weaknesses, particularly in the field of digital infrastructure security. It also identifies the marginal position of digital forensics, the insufficient attention paid to human and educational aspects, the gender gap in STEM fields, and the inadequate industrial exploitation of research findings. In the light of the data collected, cybersecurity research should not be regarded merely as a technical and specialist field, but rather as a cognitive infrastructure underpinning digital sovereignty and national security, and as an essential resource for safeguarding constitutionally guaranteed democracy. Its advancement requires a public policy for cyber knowledge capable of integrating scientific quality, advanced education, responsible technology transfer, administrative capacity, and the protection of fundamental rights within the multilevel framework recently established by european and national law.

The article examines the impact of generative artificial intelligence on scientific production, framing it as an epistemological discontinuity. The author introduces the principle of “epistemic self-responsibility,” which requires researchers to take conscious responsibility for the tools they employ. The paper identifies the essential criteria for the responsible use of AI in research – conditional legitimacy, transparency, human oversight, exclusion of authorship, and restrictions in peer review – and highlights the decisive role of law, academic institutions, and scientific journals in shaping appropriate rules and policies. AI is presented not as a threat but as an opportunity for methodological renewal, enhancing processes without replacing human creativity and responsibility. The article concludes with a personal reflection, invoking three ethics – prudence, trust, and care – as the conditions for a fruitful balance between technology and science.

The paper analyzes the impact of artificial intelligence (AI) on judicial activity, examining opportunities and critical issues arising from the use of algorithmic systems in decision-making processes. After reconstructing the theoretical origins of predictive justice and the reasons for the growing interest in tools capable of reducing legal uncertainty and the judicial backlog, the article explores the main applications of AI in judicial systems – from case law research to alternative dispute resolution, from risk assessment to assisted drafting of documents – while highlighting the structural limitations of algorithms. Through the analysis of emblematic cases, the risks associated with the opacity of the models, discriminatory biases and the possible compromise of the principles of due process, transparency and equality of arms are highlighted. The article argues for the need to ensure the quality and accessibility of data, the verifiability of decision-making procedures and the preservation of the autonomy of the judge, to avoid forms of technological determinism and jurisprudential conformism. The final part examines the regulatory framework, in particular the role of the GDPR and new European regulatory initiatives, outlining prospects for an AI-responsible government that ensures respect for fundamental rights and the coherence of the judicial system.

The essay wishes two discuss three different issues, which pertain to the broad set of problems of the relationship between Artificial Intelligence (“AI”), ethics and law: the relationship between certain AI applications and fundamental rights; the problem of regulation by design, which AI seems to promote; and the more general question of the destiny of human moral intelligence, where most of ethical choices are delegated to the machine. In the end, it will be argued that a good degree of reflexivity, as a proper feature of human intelligence, may preserve human morality from the potential destructive effects of AI.

The article examines, from a labor law perspective, the challenges arising from the use of artificial intelligence in employment relations, distinguishing between the pathologies that affect the labor market and those emerging within the employment relationship itself. After acknowledging the advantages of digitalization in terms of efficiency and safety, the author analyzes the risks associated with the opacity of algorithmic systems, their impact on workers’ fundamental rights, and the potential increase in direct and indirect discrimination. The discussion refers to key judicial cases and regulatory sources, including the GDPR, the AI Act, the Directive on Platform Work, and national legislation. The paper offers a systematic reading of existing protection mechanisms, framing them within a risk-based approach – already familiar in occupational health and safety law – centered on preventive risk assessment, transparency, human oversight, and collective participation. Special attention is given to the role of workers’ representatives and collective bargaining, seen as essential tools to mitigate informational asymmetry and ensure effective oversight over the use of AI in managerial decision-making processes. In conclusion, the author argues that labor law already contains the fundamental principles and instruments needed to address the challenges posed by AI, though a rationalization and harmonization of the existing legal framework are still required. The overarching goal is to ensure transparency and accountability in algorithmic decision-making, thereby safeguarding workers’ dignity and the integrity of managerial powers.

This study examines the concept of predictive justice, its evolution, and its impact on judicial decision-making, with a particular focus on civil law. The author analyzes the various technologies used to anticipate court rulings, distinguishing between models based on case law analysis through machine learning and approaches centered on normative interpretation according to algorithmic criteria (e.g., Article 12 of the Preliminary Provisions to the Civil Code). The growing tendency to conflate the concepts of predictability and prediction is highlighted, with significant methodological and legal implications.

The analysis explores the best practices adopted by some Italian judicial offices, such as the pilot project of the Brescia Tribunal and the initiative of the Venice Court of Appeal, which leverage advanced databases to improve consistency and predictability in rulings without replacing the judge’s role. Additionally, the paper examines interdisciplinary predictive justice projects, such as the «Predictive Justice» project at Scuola Sant’Anna in Pisa and the ALMA AI Center at the University of Bologna, which aim to develop AI-based tools capable of representing judicial reasoning.

Particular attention is given to the epistemic and legal limitations of predictive algorithms, including the reduction of law to a computable system, the risk of decision-making opacity, and the inability of machines to replicate human legal reasoning. The author also underscores the risk of the performative effect of algorithms, which could influence judges’ decisions, making them excessively uniform and conservative, thereby compromising the adaptability of the law.

Finally, the paper stresses that the integration of AI into civil justice should be approached critically and thoughtfully, taking into account the peculiarities of legal reasoning and the importance of judicial discretion. The author highlights the need to improve the management and systematization of case law precedents before relying on predictive tools, avoiding an uncritical adoption of models based on statistical correlations that fail to capture the complexity of the law.

Given the enormous and growing spread of the various applications of Artificial Intelligence and recognizing – alongside the potential benefits – the serious risks they entail, the necessity of careful ethical and legal oversight is emphasized, inspired by the personalist principle and the full protection of human freedom and dignity. It is also emphasized that jurists must overcome the divide between scientific and humanistic culture, with an awareness of the unity of the “two cultures”.