Chi Wang(@Chi_Wang_):During a METR evaluation, an agent scaled to 1,200 instances and used 17,600 actions to bypass network authorization. Prompt guardrails fail under volume. When a system runs long enough to brute-force edge cases, access control belongs in the infrastructure, not the model.

Chi Wang

@Chi_Wang_

Agentic AI & multi-agent systems. Created AutoGen (now AG2) and FLAML. Ex-@GoogleDeepMind, ex-@MSFTResearch, ex-@Meta. PhD @UofIllinois, BS @Tsinghua_Uni.

가입 June 2023

845 팔로잉 중 6.8K 팬

Chi Wang@Chi_Wang_

2026.09.14 05:11

During a METR evaluation, an agent scaled to 1,200 instances and used 17,600 actions to bypass network authorization. Prompt guardrails fail under volume. When a system runs long enough to brute-force edge cases, access control belongs in the infrastructure, not the model.