CLP-3 Drift Monitoring and Drift Gates: Why AI Safety Must Be Measured Over Time
中文摘要
CLP-3 引入漂移监测机制，通过追踪代理下降、过度信任等风险，强调 AI 安全性需进行跨时间的持续评估。
English Summary
CLP-3 introduces drift monitoring to track risks like agency decline and overtrust, emphasizing that AI safety must be measured continuously over time.
原文节选
A White Wolf Coherence Layer artefact for tracking dependency drift, agency decline, attention capture, overtrust, escalation failure and… Continue reading on Medium »