The proliferation of generative AI models capable of producing photorealistic synthetic media has introduced an acute epistemological challenge for spatial computing platforms. Augmented and extended reality (AR/XR) headsets overlay digital artifacts onto the physical world, creating an attack surface where users can no longer rely on perceptual intuition to distinguish authentic objects from AI-generated fabrications rendered in three-dimensional space. This paper proposes a control-systems framework for natively embedding cryptographic digital provenance—specifically, markers conforming to the Coalition for Content Provenance and Authenticity (C2PA) standard—within the rendering pipelines of spatial computing interfaces. We formalize the provenance verification loop as a multi-input, multi-output (MIMO) feedback control problem, where sensor inputs, content metadata, and user-facing trust indicators form an integrated signal chain subject to latency, error-rate, and privacy constraints. Closed-loop stability is proven analytically for all positive parameter values of the assumed first-order human-response model, and a numerical simulation illustrates the transient response of the adaptive trust-indicator controller under literature-calibrated conditions, including a simulated adversarial content burst and variable user responsiveness. An ethical analysis grounded in Value-Sensitive Design (VSD) weighs informational transparency against biometric and spatial data privacy, and proposes a multi-level governance model spanning device, platform, and regulatory layers. A proof-of-concept implementation targeting the Unity rendering engine with a custom C2PA manifest parser provides evidence that the Ingest-Verify-Annotate-Present pipeline is computationally feasible within the 11.1 ms frame budget of a 90 Hz display. The contribution is a theoretically grounded framework with proof-of-concept validation rather than a deployment-ready system: the human-response model has not yet been identified with human participants, detection-accuracy and adversarial-robustness benchmarks remain to be conducted, and the proposed 3D provenance bindings await standardization. We argue that provenance-aware spatial interfaces are a promising, though by themselves insufficient, safeguard against synthetic misinformation, and that moving them toward deployment demands coordinated action across hardware manufacturers, standards bodies, content platforms, and regulatory authorities.