Abstract
Systems are increasingly capable of producing outputs that appear complete and actionable. These outputs are frequently evaluated as though they fully represent the processes that produced them. This paper argues that where identity is not explicit at the interaction boundary, interpretation necessarily relies on inference. Inference enables interaction but leaves residual uncertainty regarding what participated in producing that interaction.
The paper proposes identity as a structural property of interaction rather than an inferred property of behavior or presentation. It defines the interaction-boundary condition under which identity becomes relevant and examines the implications of that condition for evaluation. The contribution is not a trust model, governance framework, or implementation architecture. Instead, it establishes a structural constraint on the claims that can be supported when identity is not explicit at the interaction boundary.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Harber, Brian, "Toward a Standard of Identity in AI Interaction", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/11309