The Human Victim Defense Framework: Legal Architecture for Autonomous AI Harm

This paper presents the Human Victim Defense Framework (HVDF) — the first complete architecture addressing all five dimensions. The framework encodes ten hypotheses, nine operational mechanisms, ten coined terms, and one new professional archetype. Every mechanism activates, anchors to, or routes through existing legal infrastructure. No mechanism creates new enforcement authority. The framework is model law ready, technology-agnostic by constitutional design, and grounded in the ASI Governance Canon (non-dominance invariant, constraint-before-capability) and the Temporal Stewardship Synthetic Integrity framework (longitudinal behavioral drift as the central failure mode, witness-not-authority as the correct stewardship posture).