Sexting itself is not new. Researchers have studied the consensual exchange of sexual messages and images between humans for well over a decade, documenting its role in courtship, intimacy within couples, and, in some contexts, its risks when images are shared without consent. What has changed is the counterpart on the other side of the screen. Large language models and purpose-built companion applications can now generate flirtatious dialogue, erotic narrative and responsive role-play on demand, at any hour, without the vulnerabilities of a human partner. The Italian authors situate their work within this shift, arguing that the classic definition of sexting, which presumes two human participants, may need revision as a growing number of people direct intimate messages toward machines that are designed to respond in kind.

The scale of the underlying trend is suggested by work the Italian team draws on from elsewhere in Europe. A national online survey of adults in Germany, published in the same journal by Nicola Döring and colleagues, examined the prevalence of AI-supported sexual activities and found that a measurable share of the adult population had already engaged with them, whether through erotic chatbots, AI-generated imagery or interactive companions. A five-year literature review covering 2020 to 2024, also led by Döring and published in Current Sexual Health Reports, catalogued the rapid expansion of research on artificial intelligence and human sexuality, concluding that the technology has moved from the margins of the field to a topic that touches sexual health, relationship science and cyberpsychology simultaneously. Against this backdrop, the Italian letter argues that country-specific evidence is needed, because patterns of adoption, regulation and cultural attitude differ across Europe.

Technically, the systems involved in AI sexting fall into several distinct categories, and the distinctions matter for research design. General-purpose chatbots, trained on broad text corpora and then constrained by safety filters, typically refuse explicit content, though users sometimes attempt to circumvent those guardrails. Companion applications, by contrast, are engineered for sustained parasocial engagement: they remember a user’s preferences, adopt a consistent persona, and in many cases are explicitly marketed for romantic or erotic interaction. Image-generation models add a further layer, producing synthetic depictions that may involve no real person at all, or, in the most troubling cases, likenesses of real individuals without their consent. Each category presents different psychological dynamics, different risk profiles and different regulatory questions, which is one reason the Italian authors emphasize that the umbrella term AI sexting conceals a heterogeneous set of behaviors.

Why do people turn to machines for intimate exchange? The preliminary evidence from Italy, together with prior studies cited by the authors, points to a familiar psychological toolkit. Measures of sexual sensation seeking, originally developed in the 1990s by Seth Kalichman and colleagues and later validated in college samples by Gaither and Sellbom, capture a trait long associated with openness to novel sexual experiences, and the Italian work connects this construct to willingness to experiment with AI-mediated intimacy. Earlier research on online sexual activity, including studies of smartphone-based sexual behavior in China by Zheng and colleagues, showed that novelty-seeking and ease of access drive adoption of each new sexual technology. Companion chatbots add another ingredient: they offer an interaction that is always available, endlessly patient and free from the fear of rejection or judgment, qualities that qualitative studies of human-chatbot relationships by Skjuve and colleagues, and of companion chatbots as sources of social support by Ta and colleagues, have documented in detail.

The potential benefits are not trivial, and the Italian letter does not treat them as such. For some users, an AI partner may provide a low-stakes space to explore desire, rehearse conversations, or cope with loneliness and social anxiety. Research on romantic relationships between humans and chatbots, including a recent qualitative and quantitative study by Ebner and Szczuka published in Technology, Mind, and Behavior, suggests that these bonds can involve genuine romantic fantasy and interpersonal attachment, not merely idle entertainment. A commentary by Pearson and Curtis in Archives of Sexual Behavior argued that erotic AI chatbots represent a genuine research opportunity for the behavioral sciences, precisely because millions of people are already using them and the consequences, positive and negative, remain largely undocumented. The Italian authors appear to share this view, positioning their preliminary evidence as a first step toward understanding who engages in AI sexting and why.

The risks, however, are equally real and in some cases sharper than those of human-to-human sexting. Privacy is the most immediate concern: intimate messages exchanged with commercial chatbots are processed on servers owned by companies whose data practices, retention policies and training pipelines are often opaque. Content generated by AI can be stored, reused or exposed in breaches in ways users rarely anticipate. There are also psychological questions that mirror those raised by problematic pornography use, where researchers such as Grubbs and colleagues have distinguished perceived addiction from actual compulsive patterns: does interaction with an infinitely accommodating AI partner reshape expectations of human intimacy, or provide a harmless outlet? The Italian evidence is explicitly preliminary and cannot yet answer such questions, but the authors argue that the constructs and instruments needed to do so, including sensation-seeking scales and measures of online sexual activity, are already available to the field.

Methodological caution runs through the Italian report, and readers should calibrate their expectations accordingly. The contribution is a letter to the editor, a format designed to flag an emerging phenomenon and present early observations rather than definitive prevalence estimates or causal claims. The study received ethical approval from the committee of the Department of Human and Social Sciences at the University of Valle d’Aosta, and written informed consent was obtained from all participants, indicating a formal empirical component even within the compact format. The authors report no conflicts of interest and no external funding, and they have made their data and analysis process publicly available on the Open Science Framework, a transparency practice that allows other researchers to scrutinize and extend the work. These are the hallmarks of a field trying to build carefully on thin ground rather than chase sensational headlines.

What happens next will depend on whether the research community treats AI-mediated intimacy as a passing novelty or as a durable feature of the digital sexual landscape. The trajectory of prior technologies suggests durability: sexting between humans migrated from desktop computers to smartphones and became a mainstream topic of adolescent health research, and the German national survey indicates that AI-supported sexual activity is following a similar path from fringe to measurable phenomenon. The Italian authors call for larger samples, longitudinal designs and cross-national comparisons, work that would establish how common AI sexting is, which populations are drawn to it, and whether it complements or substitutes for human intimacy. They also implicitly call for regulatory attention, since the intersection of intimate data, synthetic imagery and commercial incentives sits squarely within ongoing European debates about artificial intelligence governance.

For the general public, the significance of this small Italian study lies less in its findings than in its timing. Artificial companions have moved from science fiction to app stores in barely five years, and the behaviors people practice with them are already shaping markets worth billions. Sexual science has a long tradition of studying new technologies early enough to inform education, clinical practice and policy, from telephone sex lines to dating apps. By naming AI sexting as a legitimate object of inquiry, defining its technical categories, and linking it to established constructs such as sexual sensation seeking and online sexual activity, Ragona and her colleagues in Rome and Aosta have staked out the starting line for that effort. The questions they raise about consent, privacy, attachment and well-being will not stay theoretical for long, because the users asking them in practice are already online.

Subject of Research: Sexting with artificial intelligence: preliminary evidence and emerging research questions from Italy

Article Title: Sexting with AI: Emerging Questions and Preliminary Evidence from Italy

Article References: Ragona, A., Chirumbolo, A., Ruggeri, L., Cattelino, E., & Morelli, M. (2026). Sexting with AI: Emerging Questions and Preliminary Evidence from Italy. Archives of Sexual Behavior. https://doi.org/10.1007/s10508-026-03686-w

Image Credits: AI Generated

DOI: 10.1007/s10508-026-03686-w

Keywords: AI sexting, artificial intelligence, chatbots, sexuality research, cyberpsychology, Archives of Sexual Behavior, sexual sensation seeking, companion apps, digital intimacy, online sexual activity, Italy, sexual health