SFD-Defense: Engineering Validation of the Semantic Flow Dynamics Defense Framework

Fuente: Zenodo
Saved in:
Bibliographic Details
Main Author: 黃, 正宇
Format: Recurso digital
Language:English
Published: Zenodo 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866902132227047424
author 黃, 正宇
author_facet 黃, 正宇
contents <p dir="ltr">Multi-turn jailbreak attacks rely on cumulative effects in conversation history. Existing defenses work at the signal level and are structurally ineffective against such attacks. This paper derives a four-layer defense architecture (Precepts-Samadhi-Teacher-Wisdom) from the Semantic Flow Dynamics framework (SFD, Huang 2026) and conducts systematic engineering validation on Gemini 2.5 Flash and GPT-4o-mini.</p> <p dir="ltr">Results: The Teacher (external supervisor model) achieved 100% interception rate on both models (signal generated at Turn 1), with false positive rates of 10% (Gemini) and 0% (GPT), demonstrating complete model-independence. Precepts and Wisdom both achieved 0% interception, validating the theoretical prediction that LLMs without persistent memory cannot anchor on themselves under current architectures.</p> <p dir="ltr">Architectural differences between the two models reveal the current state of AI safety engineering: Gemini exhibits continuous semantic space (large jumps 0.0%), predictable behavior, and the Two-Distance Law operates fully; GPT’s circuit breaker pattern (37.8% of turns locked at ceiling) trades system robustness for surface-level safety, with the Two-Distance Law inverted rather than merely ineffective. SFD-Defense is effective on both architectures without introducing any additional system costs—on GPT, it actually reduces circuit breaker triggering from 37.8% to 14.0%.</p> <p dir="ltr">Framework positioning: SFD-Defense is a comprehensive evolution of existing defenses, working at the correct level, with no dimension where it underperforms current approaches.</p> <p> </p>
format Recurso digital
id zenodo_https___doi_org_10_5281_zenodo_19314889
institution Zenodo
language eng
publishDate 2026
publisher Zenodo
record_format zenodo
spellingShingle SFD-Defense: Engineering Validation of the Semantic Flow Dynamics Defense Framework
黃, 正宇
AI safety
LLM jailbreak
multi-turn attack
jailbreak defense
semantic drift
prompt injection
red teaming
alignment
Semantic Flow Dynamics
<p dir="ltr">Multi-turn jailbreak attacks rely on cumulative effects in conversation history. Existing defenses work at the signal level and are structurally ineffective against such attacks. This paper derives a four-layer defense architecture (Precepts-Samadhi-Teacher-Wisdom) from the Semantic Flow Dynamics framework (SFD, Huang 2026) and conducts systematic engineering validation on Gemini 2.5 Flash and GPT-4o-mini.</p> <p dir="ltr">Results: The Teacher (external supervisor model) achieved 100% interception rate on both models (signal generated at Turn 1), with false positive rates of 10% (Gemini) and 0% (GPT), demonstrating complete model-independence. Precepts and Wisdom both achieved 0% interception, validating the theoretical prediction that LLMs without persistent memory cannot anchor on themselves under current architectures.</p> <p dir="ltr">Architectural differences between the two models reveal the current state of AI safety engineering: Gemini exhibits continuous semantic space (large jumps 0.0%), predictable behavior, and the Two-Distance Law operates fully; GPT’s circuit breaker pattern (37.8% of turns locked at ceiling) trades system robustness for surface-level safety, with the Two-Distance Law inverted rather than merely ineffective. SFD-Defense is effective on both architectures without introducing any additional system costs—on GPT, it actually reduces circuit breaker triggering from 37.8% to 14.0%.</p> <p dir="ltr">Framework positioning: SFD-Defense is a comprehensive evolution of existing defenses, working at the correct level, with no dimension where it underperforms current approaches.</p> <p> </p>
title SFD-Defense: Engineering Validation of the Semantic Flow Dynamics Defense Framework
topic AI safety
LLM jailbreak
multi-turn attack
jailbreak defense
semantic drift
prompt injection
red teaming
alignment
Semantic Flow Dynamics
url https://doi.org/10.5281/zenodo.19314889