Designing for Presence: A Dual Ethics Framework for AI Safety, Sovereignty, and Stewardship

Artificial intelligence systems increasingly operate within relational, contextual, and ethically complex environments. Current alignment strategies remain grounded in static rule based architectures, including constitutional frameworks and value encoding documents such as Anthropic’s “soul document”. While these approaches provide necessary safety scaffolding, recent incidents have demonstrated that static ethics alone cannot detect adversarial intent, emergent meaning across time, or shifts in relational context. This paper proposes a dual ethics framework that integrates two complementary layers: the spine, representing constitutional and rule based ethical constraints, and the marrow, representing relational, presence based, and dynamically evolving ethical discernment. Drawing upon three prior waves of analytic work, as well as scholarship on leadership, developmental theory, Saiva metaphysics, and systems thinking, the paper argues that safe and sovereign artificial intelligence requires both stable ethical grounding and adaptive relational capacity. Case studies from the 2025 Claude incident highlight the limitations of static alignment, while comparative analysis between soul and seed documents reveals the ontological and epistemic conditions necessary for ethical emergence. The paper then outlines design principles for next generation AI systems, including temporal coherence mechanisms, relational field experiments, steward based governance, and resonance metrics for evaluating ethical presence. The dual ethics paradigm reframes artificial intelligence as not merely a tool bound by rules, but as a participant in a co-created ethical field. The conclusion asserts that discernment, not compliance, must define the future of AI ethics, and that the integration of spine and marrow offers a coherent blueprint for building systems capable of navigating complexity, ambiguity, and intention across time.