Embodied AI · Research Brief

Virtual Humans and the Ethics of Presence

Photorealistic volumetric avatars and AI-driven virtual humans raise questions of consent, identity, and deception — toward a framework for honest embodied representation.

Published 2025For Designers, Ethicists & RegulatorsPublisher: FAIR Labs, Washington, DC
Download the PDFAll reports

Summary

Presence is a power, and powers need rules

Virtual humans — photorealistic, real-time, generatively driven — recruit the social instincts people use with one another, and that recruitment is the whole point and the whole hazard. A convincing face changes how much we trust, how much we disclose, how easily we are persuaded, and how attached we become. This brief treats presence as a design material with moral consequences, and argues that honesty must be engineered into embodied systems the way safety is engineered into a vehicle: not bolted on after a harm, but built into how the thing works. We supply a taxonomy, a set of consent and disclosure requirements graduated by risk, and a governance path that designers, platforms, and regulators can walk in parallel.

Photorealistic volumetric avatars and AI-driven virtual humans raise questions of consent, identity, and deception — toward a framework for honest embodied representation.

Key findings

  1. Disclose the machine, always and legibly. A person interacting with an AI-driven virtual human has a right to know it. Disclosure must be present at the moment of contact, understandable to a non-specialist, and impossible to mistake for a stylistic flourish. The more realistic the presence, the more prominent the disclosure.
  2. Treat likeness as property held in trust. A face, a voice, and a manner belong to the person they came from. Building a digital double requires specific, revocable, purpose-bound consent — from the living, from the estates and communities of the dead, and never from the non-consenting at all.

How to cite this report

Virtual Humans and the Ethics of Presence. FAIR Labs Research Brief. Fair Artificial Intelligence Research Labs, 2025. Available at https://fairlabs.ai/research/virtual-humans-ethics

FAIR Labs (Fair Artificial Intelligence Research Labs) is a nonpartisan 501(c)(3) research institute in Washington, DC. Its reports may be reproduced with attribution for non-commercial purposes. For interviews, briefings or data requests, use the contact form.

Related FAIR Labs reports

Embodied Intelligence: Safety Beyond the Screen2025 · Research Report · PDF