Year: 2026
Type: article
Abstract: Smart glasses can infer emotion, intention, and vulnerability from facial, vocal, and physiological signals. This paper argues that computationally reading people can erode the very signals it seeks to read. Where interaction must continue and exit is costly, asymmetric inference makes concealment rational: masks, adaptive lenses, flattened expression, counter-analysis, and AI-mediated exchange. Each defence reduces legibility, and each removes nuance from practice. The paper names this adversarial cycle nuance warfare and states the conditions under which it emerges. Where those conditions coincide, both parties would end up concealing, an equilibrium the paper calls mutual assured opacity, in which interaction continues through reduced signals while natural nuance is withdrawn. Sustained withdrawal can narrow the settings in which nuance is practised and, where machine mediation becomes routine before independent practice develops, leave later cohorts without the experiential substrate that reading another person requires. Against this, the paper sets out a constructed alternative, mutual assured transparency, in which parties disarm their instruments at an agreed threshold, and proposes the right to remain unread: protection against the machine-mediated extraction of strategically actionable inferences from involuntary or contextually unavoidable human signals. Existing research on adversarial face recognition, on mask-related losses in machine and human emotion recognition, and on channel switching supports the opening transitions; the later escalation is offered as conceptual foresight grounded in structural analogy, and the paper states the settings in which it is not predicted.
Source: Discover Informatics
Author: Kenji Yamada
Language: English
Subfield: Safety Research
Field: Social Sciences
Domain: Social Sciences
Open Access status: hybrid
OpenAlex ID: W7223137128