Where extended reality and AI may take us: Ethical issues of impersonation and AI fakes in social virtual reality

In plain English

At a panel held inside social VR at IEEE VR 2025, where each panellist appeared as a realistic avatar of themselves, two panellists secretly swapped avatars and tried to act as each other; roughly 40 of 100 surveyed audience members never noticed. An LLM-driven avatar modelled on Alan Turing also joined the panel and was rated less realistic than the humans and somewhat distracting. For anyone deploying photorealistic avatars, the lesson is that viewers cannot be relied on to detect identity substitution, so authentication and provenance signals have to come from the platform rather than from the audience's judgement — and the authors note they never sought permission to depict Turing, which is its own cautionary detail.