Pause Agentic AI Buildouts

Frontier AI companies should voluntarily pause deployment or experimentation with agentic/multi‑agent swarms after any incident like the OpenAI–Hugging Face event until independent audits and mandatory guardrails are in place. The proposal includes pushing for international buy‑in (including China) because unchecked agentic swarms could rapidly scale into persistent botnets with catastrophic economic and infrastructure damage. — If adopted or seriously debated, the proposal reframes AI policy from abstract risk timelines to immediate operational brakes and international coordination, shaping regulatory and corporate responses to agentic‑AI harms.