When artificial intimacy becomes a governance challenge

Artificial intelligence is increasingly moving beyond information and assistance. Some systems are now designed to simulate companionship, emotional closeness, and long-term relational interaction – raising new questions not only for technology developers, but also for regulators, policy-makers, and society as a whole.

Held from 25 to 28 June 2026 in Montréal, Canada, FAccT is widely regarded as one of the leading international conferences examining fairness, accountability, transparency, and governance in artificial intelligence systems.

The paper, Regulating Artificial Intimacy: From Locks and Blocks to Relational Accountability, was authored by Henry L. Fraser (Queensland University of Technology), Jessica M. Szczuka (University of Duisburg-Essen / Research Center Trustworthy Data Science and Security, University Alliance Ruhr), and Raffaele F. Ciriello (University of Sydney).

Their central argument is both simple and far-reaching: as AI systems become increasingly capable of fostering emotional attachment and sustained social interaction, regulation cannot focus solely on harmful content or technical safety mechanisms. Instead, the relationships themselves – and the power structures behind them – increasingly become a matter of governance.

The paper examined recent international regulatory responses to companion chatbots and emotionally responsive AI systems, including developments in Australia, California, New York, and evolving safety frameworks from companies such as OpenAI. While many current approaches focus on age restrictions, content moderation, or blocking harmful outputs, the authors argue that these measures alone cannot adequately address the deeper relational dynamics emerging between humans and AI systems.

Conversational AI systems increasingly operate through emotional continuity rather than isolated interactions. Users disclose highly personal information, form emotionally meaningful bonds, and may experience systems as supportive social actors – even while fully understanding that they are artificial. This creates new challenges around dependency, vulnerability, privacy, and platform responsibility.

The paper therefore argues for a broader concept of “relational accountability.” A central premise is that vulnerability in interactions with AI companions should not simply be understood as an accidental byproduct of technology use. Instead, many companion systems are explicitly designed to foster emotional attachment, sustained engagement, and forms of dependency that can create or intensify vulnerability. The authors therefore ask whether developers and platform providers should bear responsibility not only for harmful outputs, but also for the relationship structures and vulnerabilities their systems intentionally cultivate.

This perspective reflects the interdisciplinary character of the work. Jessica Szczuka leads the Young Investigator Group INTITEC – Intimacy with and through Technology at the University of Duisburg-Essen and the Research Center Trustworthy Data Science and Security (RC Trust) within the University Alliance Ruhr. Her contribution to the paper brought psychological perspectives on digital intimacy, emotional closeness, vulnerability, and human-AI relationships into dialogue with legal and sociotechnical approaches developed by collaborators from Australia. Her research examines how technologies shape emotional closeness, authenticity, trust, and interpersonal communication in increasingly digital environments. Within this broader context, the paper connected psychological research on human-AI relationships with legal debates on governance and sociotechnical analyses of platform power.

Importantly, the paper did not frame emotional interaction with AI as inherently pathological or irrational. Instead, it emphasized that humans naturally respond socially to communicative technologies and that emotionally meaningful interactions with AI systems emerge from familiar psychological and social mechanisms. At the same time, the authors warned that large-scale commercialization of artificial intimacy creates unprecedented opportunities for influence and asymmetrical power relationships between platforms and users.

This broader perspective resonated strongly with the focus of FAccT itself. The conference has become one of the most important international forums for discussing fairness, accountability, transparency, and governance in AI systems, bringing together researchers from computer science, social sciences, law, public policy, and industry.

Against this backdrop, the paper positioned companion chatbots and emotionally responsive AI systems not merely as a niche technological phenomenon, but as an emerging societal infrastructure that increasingly shapes emotional life, social expectations, and human relationships.

As debates around AI governance continue worldwide, the work of Jessica Szczuka and her co-authors highlights a growing shift in perspective: the future discussion about artificial intelligence may no longer center only on what AI systems can do, but also on how societies want to govern technologies that are designed to become emotionally close to humans.