Ethics & Misuse This work analyzes structural influence mechanisms (RCM×TCM×Class) that may operate without explicit falsehood, through composition control (DL/IL/RL), boundary spotlighting, and verification bypass (“blur engineering”). The author acknowledges that merely recognizing such mechanisms can imply potential misuse. For this reason, the manuscript intentionally excludes operational manipulation instructions, tactical scripts, or step-by-step influence procedures. The intended use of this work is strictly limited to public-interest defense, safety analysis, verification protocols, and institutional evaluation. Any attempt to optimize real-world influence, political persuasion, or coercive manipulation using this framework is explicitly outside scope and ethically prohibited. This paper proposes a unified framework (RCM×TCM) under the assumption that influence operations do not necessarily require falsehood and can be achieved even with factually true statements. The Reality Composition Model (RCM) conceptualizes perceived reality as a composition of Direct Light (DL), Indirect Light (IL), and Reflected Light (RL), and argues that manipulation can arise via emphasis, omission, and weight shifting among these components. The Threshold Capture Model (TCM) posits that policy/system boundaries (thresholds) and the Threshold Class (T) constitute a core capture surface. The paper further extends target taxonomy into BAC (Belief-Anchor Class), LIC (Low-Involvement Class), and T (Threshold Class), showing that identical compositional operations yield different outcomes depending on group characteristics. Blur is defined as a verification-bypass design layer encompassing cushioning, delay, ambiguity, dilution, responsibility shifting, and session fragmentation. The paper presents structural detection rules and the KEI-VX defense protocol, and includes a reproducible evaluation/verification plan. “This record includes KEI × AI Collaboration Charter v1.0 (ethics & non-weaponization).” 본 논문은 영향력 조작(influence)이 반드시 허위(falsehood)를 필요로 하지 않으며 사실 기반으로도 성립할 수 있다는 전제 하에 통합 프레임워크 RCM×TCM를 제안한다. RCM(Reality Composition Model)은 현실 인식이 직접광(DL), 간접광(IL), 반사광(RL)의 구성으로 형성됨을 전제하고, 조작은 이 구성요소의 강조·삭제·가중치 조절을 통해 발생할 수 있음을 설명한다. TCM(Threshold Capture Model)은 정책·제도·시스템의 경계면(Threshold)과 임계층(T)이 영향력 작동의 핵심 지형이 됨을 주장한다. 또한 본 논문은 대상군을 BAC(신념 확고층), LIC(저관여층), T(임계층)으로 확장 분류하여, 동일한 구성 조절이라도 집단 특성에 따라 결과가 달라짐을 보인다. 흐림(blur)은 완충·지연·모호화·희석·책임 전가 및 세션 파편화를 포함하는 검증 우회(verification bypass) 설계로 정의되며, 마케팅·정치·AI 상호작용에서 공통적으로 관찰되는 구조를 통합 모델로 제시한다. 최종적으로 본 논문은 구조 기반 탐지 규칙과 방어 프로토콜(KEI-VX)을 제시하고 재현 가능한 평가·검증 설계를 포함한다.