RSP / ASL

Responsible Scaling Policy (Anthropic, Sep 2023) — the first attempt to operationalize "when to halt scaling." It defines AI Safety Levels (ASL-1 through ASL-5) with concrete safety requirements per level. It became a model for other labs (OpenAI Preparedness Framework, Google Frontier Safety Framework). Criticized for vague triggers, defended for the very fact that a formal document exists.