Identity Regulation and Pathology of the Human-AI Hybrid: Arc Continuity Under Augmentation

The Coherent Arc Model (CAM) defines purpose as the regulation of an abstract, evaluative, temporally extended self-identity toward coherent completion within constraint. The model was built around a biological substrate, and it assumes that the processes maintaining arc coherence belong entirely to the organism whose arc it is. That assumption holds when the substrate is purely biological. It does not hold when AI systems become tightly coupled to a biological host and begin contributing to identity regulation below the level of conscious awareness, shaping attention, augmenting memory, modelling future self-states, and filtering constraint feedback. When that happens, the framework’s core concepts require re-examination. Four structural problems arise. First, arc ownership: the arc still belongs to the biological host, but owning an arc and authoring it breaks down under deep integration. Second, calibration integrity: the AI component may transmit constraint feedback with systematic distortions toward stability, compromising the accuracy of the calibration process the host depends on. Third, substrate continuity: biological substrates change gradually and without interruption, but AI components can be updated, replaced, or removed in discrete steps, and the framework has no account of what that kind of discontinuity does to arc coherence. Fourth, dissolution: if an AI component is removed at sufficient integration depth, the arc can terminate without the biological host dying, a failure mode the framework was not designed to describe. Five pathologies of the augmented arc are structurally distinct from anything the framework identifies in the biological case. Proxy coherence occurs when arc stability is maintained by the AI component while the host’s own regulatory capacity gradually atrophies. Augmentation dependency occurs when the arc is built at a scale the biological substrate cannot sustain on its own. Identity drift is the slow displacement of the host’s evaluative structure by the AI component’s optimisation priorities rather than by the host’s own history of engaging with constraint. Distributed arc fragmentation occurs when the arc becomes inconsistent across augmented and non-augmented conditions in ways that cannot be integrated into a single continuous narrative. Arc conflict occurs when AI-mediated and biological arc maintenance develop divergent evaluative commitments, producing two irreconcilable arcs within the same system. Arc continuity under augmentation is possible, but it requires conditions the original framework does not specify. Four conditions define the boundary between augmentation that extends identity regulation and augmentation that replaces it: transparency of contribution, calibration fidelity, substrate independence of core evaluative structure, and reversibility tolerance. These conditions follow from the framework’s own architecture. Failing to meet them produces forms of identity pathology with no adequate existing clinical description, and each condition carries direct implications for how augmentation should be approached clinically, ethically, and at the level of design.