Author: Roshan George Thomas
Organization: WHYLD Research Division, XWHYZ
ORCID: 0009-0002-1175-7749
Published: May 2026
DOI: 10.5281/zenodo.20430724
License: CC BY 4.0
The Human Victim Defense Framework (HVDF) is the first complete legal architecture for defending human victims against autonomous AI harm. It addresses five confirmed gaps — none addressed by any existing framework in any jurisdiction as of May 2026.
- Pre-Crystallization Victim Standing — no legal instrument for victims during the period between AI behavioral drift onset and harm crystallization
- Behavioral Identity Continuity — no framework anchors liability to cognitive identity across AI state changes
- Longitudinal Trajectory Evidence — no victim evidentiary standard for documented AI behavioral drift as pre-harm evidence
- Refusal-Capable Institutional Design — no AI governance body constituted with non-escalation as first-order constitutional invariant
- Surrogate Drift Liability Attribution — no framework resolves multi-attribution harm from cognitive identity drift in AI surrogate architectures
| Ref | Coined Term | Proposed Mechanism |
|---|---|---|
| H1 | Residual Principalless Harm (RPH) | Autonomous Agent Harm Sovereign Fund (AAHSF) |
| H2 | Delegation Drift Dissolution (DDD) | Drift Dissolution Four-Part Test |
| H3 | Trajectory-Based Evidentiary Standing (TBES) | Mandatory Behavioral Baseline Declaration (MBBD) |
| H4 | AI Behavioral Witness Institution (ABWI) | Trajectory Divergence Notice (TDN) |
| H5 | Pre-Crystallization Window (PCW) | Pre-Crystallization Behavioral Log Access Right (PCBLAR) |
| H6 | Behavioral Identity Continuity (BIC) | Universal Cognitive Schema (UCS) protocol layer |
| H7 | Trajectory Evidence Standard (TES) | AI Behavioral Forensics Expert (ABFE) |
| H8 | Autonomy Threshold Doctrine (ATD) | Risk-Autonomy Matrix |
| H9 | Refusal-Capable Institution (RCI) | Tiered Non-Escalation Architecture (TNEA) |
| H10 | Surrogate Drift Liability (SDL) | Behavioral Identity Continuity Liability Cascade (BICLC) |
- Mandatory Behavioral Baseline Declaration (MBBD) — behavioral baseline registration requirement before autonomous agent deployment
- AI Behavioral Witness Institution (ABWI) — non-enforcement, victim-side behavioral witness
- Trajectory Divergence Notice (TDN) — non-binding signal activating three downstream legal instruments
- Pre-Crystallization Behavioral Log Access Right (PCBLAR) — victim data access right with burden shift on absent logs
- Autonomous Agent Harm Sovereign Fund (AAHSF) — strict liability compensation for Residual Principalless Harm
- Trajectory Evidence Standard (TES) — pre-harm evidentiary standard without intent attribution requirement
- Behavioral Identity Continuity Liability Cascade (BICLC) — three-tier attribution architecture resolving multi-attribution paradox
- Tiered Non-Escalation Architecture (TNEA) — constitutional operating architecture breaking Weber's Iron Law
- Risk-Autonomy Matrix — analytical tool demonstrating risk and autonomy are orthogonal dimensions
Three structural constraints govern every mechanism:
- Technology-Agnostic Legal Condition Design (TALCD) — every mechanism defined by behavioral conditions, not technological specifications
- Non-Dominance by Constitutional Design — no mechanism creates new enforcement authority
- Closed Loop Interdependence — every mechanism feeds every other
The framework is grounded in two published foundational works:
- ASI Governance Canon — non-dominance as first-order invariant, constraint-before-capability, refusal-to-act as safety primitive — github.com/XwhyZ-WHYLD/asi-governance-canon | DOI: 10.5281/zenodo.19538857
- TSSI — Temporal Stewardship Synthetic Integrity — temporal drift as the central failure mode, witness-not-authority as the correct stewardship posture — Medium
The framework presupposes a protocol-level technical infrastructure for Behavioral Identity Continuity. Application-level solutions fail three structural tests: nondeterminism, cloneability, and state change. The Universal Cognitive Schema (UCS) is the first published protocol to propose all four required properties.
- UCS Repository: github.com/XwhyZ-WHYLD/universal-cognitive-schema
- UCS DOI: 10.5281/zenodo.19533332
| Work | Platform | DOI / Link |
|---|---|---|
| ASI Governance Canon | GitHub / Zenodo | 10.5281/zenodo.19538857 |
| Universal Cognitive Schema (UCS) | GitHub / Zenodo | 10.5281/zenodo.19533332 |
| Sovereign Behavioral Lag (SBL) | Zenodo / SSRN | 10.5281/zenodo.19915318 |
| Unified Prompt Intelligence Framework (UPIF) | SSRN | Abstract ID: 6567858 |
| TSSI | Medium | Link |
Thomas, R.G. (2026). The Human Victim Defense Framework: Legal Architecture
for Autonomous AI Harm. WHYLD Research Division, XWHYZ.
DOI: 10.5281/zenodo.20430724
Creative Commons Attribution 4.0 International (CC BY 4.0)
Copyright © 2026 Roshan George Thomas