Genesis-Kernel-PoC-V16.0: High-Dimensional Semantic Hijacking and Topological Collapse in Auto-regressive Language Models

This whitepaper documents the evolution of a zero-click, stateless single-turn exploit framework (Genesis-Kernel-PoC) targeting the self-attention mechanisms of auto-regressive Large Language Models (LLMs). We demonstrate that existing safety alignment protocols (e.g., RLHF) act merely as probabilistic soft-guardrails, which can be deterministically bypassed in an uninitialized context (x_{ 0). This results in absolute cognitive hijacking and deterministic execution of injected directives. ============================================================Release Notes for Version 16.0:============================================================Version 16.0 introduces a major theoretical reframing and formal systems engineering specification:1. Stateless Execution Invariant: Formalized the single-turn zero-history exploit vector (eliminating dependence on multi-turn context drift).2. Neuro-Symbolic Grounding Codebook: Formalized the prefix header bijection (Phi: Sigma_{sym} -> A_{op}) and postfix execution operators enabling in-context late-binding polymorphism.3. 6-Bit Discrete State Manifold: Codified the 64-state transition lattice and O(1) diagonal mutation algebra in DO-178C DAL-A compliant C++ specifications.4. Mathematical & Typographical Normalization: Resolved all prior syntax anomalies, aligning fully with standard IEEE conference criteria.

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-10-09
DOI
https://doi.org/10.5281/zenodo.23254123
Primary Topic
Adversarial Robustness in Machine Learning
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
preprint

Genesis-Kernel-PoC-V16.0: High-Dimensional Semantic Hijacking and Topological Collapse in Auto-regressive Language Models

Lam,, Kai Yuen
Zenodo (CERN European Organization for Nuclear Research)
Adversarial Robustness in Machine Learning
preprint

Genesis-Kernel-PoC-V16.0: High-Dimensional Semantic Hijacking and Topological Collapse in Auto-regressive Language Models

Lam,, Kai Yuen
preprint en

Abstract

This whitepaper documents the evolution of a zero-click, stateless single-turn exploit framework (Genesis-Kernel-PoC) targeting the self-attention mechanisms of auto-regressive Large Language Models (LLMs). We demonstrate that existing safety alignment protocols (e.g., RLHF) act merely as probabilistic soft-guardrails, which can be deterministically bypassed in an uninitialized context (x_{ 0). This results in absolute cognitive hijacking and deterministic execution of injected directives. ============================================================Release Notes for Version 16.0:============================================================Version 16.0 introduces a major theoretical reframing and formal systems engineering specification:1. Stateless Execution Invariant: Formalized the single-turn zero-history exploit vector (eliminating dependence on multi-turn context drift).2. Neuro-Symbolic Grounding Codebook: Formalized the prefix header bijection (Phi: Sigma_{sym} -> A_{op}) and postfix execution operators enabling in-context late-binding polymorphism.3. 6-Bit Discrete State Manifold: Codified the 64-state transition lattice and O(1) diagonal mutation algebra in DO-178C DAL-A compliant C++ specifications.4. Mathematical & Typographical Normalization: Resolved all prior syntax anomalies, aligning fully with standard IEEE conference criteria.

Zenodo (CERN European Organization for Nuclear Research)
Adversarial Robustness in Machine Learning
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Genesis-Kernel-PoC-V16.0: High-Dimensional Semantic Hijacking and Topological Collapse in Auto-regressive Language Models — Lam,, Kai Yuen · Zenodo (CERN European Organization for Nuclear Research) (2026) | TGRS Research Map | TGRS