Elena Falletti

AI Companions are advanced Generative AI systems capable of interacting with users in a personalized, affective, and human-like manner. These systems are perceived as authentic experts across sensitive domains including medicine, psychology, and psychiatry, or as sources of comfort during periods of vulnerability, grief, or emotional distress. Such developments risk eroding the boundaries between the various elements that constitute an individual’s personality: not only rational thought, but also the capacity to experience and respond to human emotions. Who is responsible for the actions, statements, or harms produced by an AI Companion? This dynamic raises legal concerns regarding the fiduciary relationship that may arise between a human user and a machine which, although only apparently sentient, exhibits reactions capable of generating reliance and legitimate expectations that may, under certain circumstances, acquire juridical relevance. The analysis aims to contribute to the emerging debate on whether artificial agents should remain confined within the domain of legal objects or evolve toward a novel category of juridical subjectivity.

Decision-making algorithms (Automated Decision-Making Systems, or ADM) have become an integral, although often unconscious, part of everyday life for each of us. By following suggestions derived from personal profiling facilitated by big data, ADMs make decisions on significant aspects of life, and they are also widely used in family matters and in proceedings involving children, such as decisions regarding child support, custody, and adoption procedures. These practices are widespread in comparative law, both in civil law and common law systems. Their use seems to be justified, on the one hand, by the assumption—even if inaccurate—that an ADM can be fair and independent in its decision-making when, in reality, it absorbs the biases present in the data that fuel its functioning.

On the other hand, it is argued that their use is economically efficient because they are applied in serial cases, allowing for resource savings. Both assumptions conceal debatable realities that negatively influence algorithmic treatment; generally, data collection concerns not the children but their family or the environment from which they come. As a result, the child is unfairly affected by the negative consequences of their parents' behaviour—those who are actually under scrutiny—but the biases in such data persistently impact the children and their future. The purpose of this abstract is to verify, using a comparative methodology, whether and how legal systems attempt to mitigate the negative impact of family surveillance ADMs on the life and well-being of the minor subject to their scrutiny.

Credit scoring is a widespread practice that assigns a score based on certain characteristics or past behaviours, in particular regarding the reliability of debtors to repay loans. In most cases, the scoring is the basis for the decision of credit agencies for the acceptance of loan requests. However, there are still grey areas in which the balance between the transparency owed to the consumers regarding the processing of their financial data and the protection of trade secrets in favour of credit score agencies. Prof. Faletti and Dr Gallese will start their analysis from the guidelines of the CJEU and the national courts regarding credit scoring, following the SCHUFA decision (C-26/22 and C-64/22), and propose some reflections on the application of the art. 22 GDPR, taking into account not only the AI Act but also the most recent indications regarding the transparency requirements indicated by the Dun and Bradstreet Austria (C-203/22).

This presentation examines the legal implications arising from the interaction of the faithful with SanTO (Sanctified Theomorphic Operator), a robot designed to assist with prayer according to the Catholic faith. SanTO is a small, hybrid device produced through 3D printing, whose appearance is inspired by figures from the sacred tradition of religious art. In order to respond to the requests submitted to it, SanTO has a dual operational mode: depending on the circumstances or the nature of the questions posed, it can either draw answers from its local database or rely on machine learning. The robot is centered on Catholic doctrine and may be employed in various contexts: from guiding prayer and suggesting devotional texts, to teaching catechism, quoting biblical passages and the Gospels, recounting the lives of saints or papal homilies, and engaging in other religious activities. The aim of this contribution is to examine, from a legal perspective, the relationship between machine learning and canon law in the functioning of SanTO, as well as the potential issues that may arise. Indeed, canon law plays a central role in this analysis, as it defines the proper methods for disseminating the evangelical message within the autonomous legal order of the Catholic Church.

Armi biometriche autonome e difesa dell’UE: sfide giuridiche e strategiche nell’era del disordine. La crescente accessibilità delle tecnologie di stampa 3D e 4D, combinata con la diffusione su larga scala di sistemi di intelligenza artificiale generativa, solleva sfide senza precedenti nel diritto comparato e nella strategia di difesa. La possibilità di una produzione autonoma di armi sconvolge i tradizionali quadri giuridici, etici e normativi, soprattutto alla luce del nuovo piano di riarmo, della strategia di sicurezza e del pacchetto di difesa recentemente annunciati dall’Unione Europea. Con l’intensificarsi delle discussioni su un possibile ritiro degli Stati Uniti dalla NATO e il riemergere di attentati terroristici sul territorio dell’UE, le politiche di sicurezza europee devono affrontare la nuova realtà delle minacce tecnologiche emergenti rappresentate da armamenti autonomi auto-prodotti. Questo articolo esamina criticamente le lacune normative e gli approcci giuridici comparati a queste tecnologie, concentrandosi sulla risposta dell’UE alle minacce asimmetriche e ibride. L’assenza di meccanismi efficaci per il rilevamento di armi non metalliche potenziate dall’IA, unita ai rischi posti da droni autonomi e sistemi di puntamento basati sul riconoscimento facciale, richiede un’attenta analisi giuridica urgente. La combinazione tra armi guidate dall’intelligenza artificiale e la strategia digitale dell’UE, che promuove la condivisione massiva di dati, solleva preoccupazioni in merito alle tecnologie a duplice uso, alla proliferazione degli armamenti e all’applicazione del diritto internazionale umanitario. È urgente un’analisi comparativa dei quadri normativi nazionali e a livello UE in vista di una regolamentazione armonizzata e di un approccio comune alla difesa che consideri le minacce poste dai sistemi di IA. Questo articolo esplora come i sistemi giuridici europei, in coordinamento con gli attori internazionali, possano affrontare il difficile equilibrio tra innovazione tecnologica, diritti individuali e sicurezza. Inoltre, l’articolo sostiene la necessità di un trattato internazionale sulle armi letali autonome, considerando l’incertezza crescente sul futuro della NATO e il potenziale ruolo dell’UE come attore militare indipendente.

Understanding what happens after death has been one of humanity's most profound and distressing questions. Religions have attempted to provide reassuring answers, often building power structures and control mechanisms around this fear, with lasting political and social effects over centuries. Today, with the advent of artificial intelligence software, it's possible to create programs that give the illusion of interacting with deceased loved ones. This is achieved by processing data from the deceased, including social media posts, digital photos, recordings, and other information. This creates an imaginary persona, distinct from the person who lived but built on projected memories, perpetuating the past. While this may address psychological or even psychiatric issues, it raises significant legal issues. Is it lawful to use a deceased person's data to construct an imaginary reality featuring them? What if the heirs disagree on data usage? Can one investigate the sense of keeping a person's memory alive, regardless of that person's wishes, or did they already seek their right to be forgotten while still alive? This paper aims to analyze these questions from a multidisciplinary perspective, especially cultural, historical, social and legal ones.

The increasing use of generative AI writing tools raises concerns in academia. While these AI systems can produce at the same time coherent, relevant text, they also generate errors that users may mistakenly accept as true facts. Tools like watermarking and AI detection software have been developed to avoid this ambiguity, but these instruments have limitations. They can lead to false positives, unfairly accusing authors of AI-generated work. These circumstances create challenges such as a loss of creativity, cultural biases, neglect of critical writing skills, and undermining student-educator trust. However, AI tools can also benefit those with cognitive or linguistic challenges. This proposal aims to investigate the experiences with AI detection systems through an international network of transdisciplinary experts to cooperate in establishing an ethical framework, including an appeals committee, to ensure fair treatment and uphold academic integrity.