Ethical Hyper-Velocity (EHV): доказуемо детерминированная архитектура JIT-компилятора, учитывающего управление, для агентных систем

Аннотация

随着自主代理系统在受监管的关键基础设施中规模化部署，缺乏基于硬件的机械性机制来实施高频策略更新，构成了根本性的安全缺口。本文提出伦理超速（EHV）——一种新型架构框架，用于在运行时对人工智能治理策略进行形式化验证。与引入14-30天延迟的追溯性审计框架（如ISO/IEC 42001、NIST AI RMF）不同，EHV通过治理感知即时编译器将策略执行点（PEP）迁移至推理流水线。通过集成无冲突复制数据类型（CRDT）实现策略同步，并在可信执行环境（TEE）内采用基于周期的证明缓存，EHV实现了亚毫秒级形式确定性（SMFD）。我们通过TLA+形式化验证证明，在该系统有界运行状态空间中，不合规的代理行为在计算上不可达。我们同时证明，O(1)运行时执行能够消除部署速度与治理完整性之间的传统权衡，将治理延迟从O(天)级降至O(1)级。

English

As autonomous agentic systems scale across regulated critical infrastructures, the lack of mechanistic, hardware-rooted enforcement for high-frequency policy updates presents a fundamental safety gap. We introduce Ethical Hyper-Velocity (EHV), a novel architectural framework for the formal verification of AI governance policies at runtime. Unlike retrospective auditing frameworks (ISO/IEC 42001, NIST AI RMF) which introduce 14-30 day latencies, EHV relocates the Policy Enforcement Point (PEP) into the inference pipeline via a Governance-Aware Just-In-Time (JIT) Compiler. By integrating Conflict-free Replicated Data Types (CRDTs) for policy synchronization and Epoch-based Attestation Caching within Trusted Execution Environments (TEEs), EHV achieves Sub-millisecond Formal Determinism (SMFD). We demonstrate via TLA+ formal verification that non-compliant agentic actions are computationally unreachable within the system's bounded operating state space. We prove that O(1) runtime enforcement can eliminate the traditional trade-off between deployment velocity and governance integrity, reducing Governance Latency from O(days) to O(1).