Project META-TOO and ethics of XR

The most recent webinar organized by our Thematic Working Group 4 (TWG4) on intersectional approaches brought together two perspectives on a perpetual problem in mainstream technology policy: harassment in virtual reality environments and the ethical frameworks needed to address it before harm becomes the norm.

A project built around a difficult question

Project META-TOO, a European research initiative coordinated by the National and Kapodistrian University of Athens (NKUA) with partners INRIA and IDIBAPS Barcelona, examines gender-based inappropriate social interaction in social VR environments. Michele Barbier, ethics specialist at INRIA, presented the project’s approach and its current findings.

Her starting position is that harassment in virtual environments is not a softer version of online abuse. Because social VR involves synchronous voice, body tracking, and physical co-presence through embodied avatars, the psychological impact of harassment registers with the same intensity as harassment in physical space. Research collected by the META-TOO team documents cases that participants and researchers have begun describing as “meta-rape”, a term that reflects both the severity of the experience and the inadequacy of existing frameworks to address it. Women, gender minorities, LGBTQIA+ users, and racialised minorities are disproportionately targeted.

The safety tools currently available on social VR platforms, such as proximity controls, mute and block functions, reporting mechanisms, and friend-tagging systems, are fragmented and inconsistent, and most of them activate only after harm has already occurred. Few are grounded in behavioural theory or evaluated comprehensively. META-TOO’s research responds to this gap through two parallel studies.

• The first, led by INRIA, focuses on safety mechanism prototyping, developing a new characterisation framework for inappropriate behaviour and testing consent-based interaction and human-AI collaborative moderation through a joint experimentation platform.
• The second, led by IDIBAPS, examines the emotional and psychological impact of harassment by varying avatar self-similarity, harassment modality and intensity, and the presence of AI agents to measure how scenario design influences embodied emotional response.

One of the more important elements of the project’s methodology is its ethics advisory board, assembled specifically to address the paradox at the centre of the research: studying harassment requires exposing participants to harassment scenarios. Managing that paradox (how far to go, what information is worth what cost) has required the involvement of computer scientists, psychologists, philosophers, and digital law jurists working in parallel. Community co-design sessions with LGBTQIA+ groups have been used to ensure that experimental avatars do not reproduce the stereotypes the research is trying to critique. A shared glossary is currently being developed across the consortium, as the first step toward a common framework that can cross disciplinary and institutional boundaries. A white paper with actionable recommendations for developers, policymakers and the industry is planned for spring 2027.

Technology is never neutral

Mariachiara Tirinzoni, philosopher and member of TWG4, offered the second half of the webinar as a framework for thinking about why technical safety tools alone are insufficient.

Her argument rests on a claim that is simple but consequential: technology is not a neutral tool. Every artifact carries within it the assumptions, cultural biases, political choices, and economic goals of those who made it, and that is a structural feature of how human beings make things. The concept of affordance, borrowed from psychology and design theory, captures this: objects invite certain uses and discourage others through their design, and those invitations are never arbitrary.

This becomes a governance problem when the defaults embedded in technology systematically advantage some users over others. Tirinzoni developed this through the metaphor of the standard, the flag placed on a battlefield to call troops together in an emergency. Standards are necessary, but they are not neutral. Those who match the standard most closely are most protected. Those furthest from it are most exposed. And because no individual perfectly matches any standard, what we are really measuring is degrees of distance from a definition that was itself a choice.

Design needs diversity and, for the same reason, it needs standards. The standard provides a basis for comparison and quality control. Diversity prevents the standard from becoming a cage that excludes the people the technology was never designed to serve. Holding both simultaneously is not contradictory and is the actual work of ethical design.

The tool Tirinzoni proposed for doing that work is moral imagination, or the capacity, which philosophers from John Dewey to Martha Nussbaum have argued is both trainable and civic, to place oneself in another’s situation before deciding. Dewey described moral deliberation as a rehearsal of possibility, akin to theatrical rehearsal, in which we imagine the consequences of choices before we make them. Nussbaum frames the ability to read stories and see oneself as an audience to events beyond one’s own experience as a civic competence, not a personal virtue. If technology encodes the intentions of those who design, deploy, and use it, then what technology we get depends in part on how imaginatively those people can conceive of lives and situations unlike their own.

The closing observations from both speakers share the belief that questions, at this stage, matter more than the answers. A common glossary, an ethics advisory board, a philosophical framework that asks about the kind of persons designers are choosing to become through the choices they make.