Working Group on AI Pause Feasibility on AI

Working Group on AI Pause Feasibility

Interdisciplinary academic working group organized by UC Berkeley scholars William Fithian and Wesley H. Holliday that studies the feasibility of mutually verifiable pauses of frontier AI training.

A hardwired pause would require participating states to [...] conduct a global census to account for the pre-pause stock of training-capable AI chips, and govern all training-capable chips. [...] Each hardware system under treaty governance could be placed in the care of a licensed hardware operator at a known data center. Core licensing, registration, and tracking requirements would apply to all accelerators under treaty governance.

AI Verified (Oct 10, 2026)

Source:

Working Group on AI Pause Feasibility, “The Feasibility of a Hardwired Pause of Frontier AI Training,” version 1.0.1, October 10, 2026. The group’s proposal would require states to census and govern all training-capable AI chips and apply licensing, registration, and tracking to accelerators under treaty governance, with each hardware system held by a licensed operator at a known data center.

20d ago

Quote authenticity verification history

Report this

AI Verified The Working Group’s Oct. 10 paper contains the quoted hardwired-pause requirements (lines 249–257; 607). · Hector Perez Arenas gpt-5.6-sol · 20d ago

replying to Working Group on AI Pause Feasibility

Other opinions from Working Group on AI Pause Feasibility

See all

Working Group on AI Pause Feasibility votes For

Require all advanced AI chips to be registered and traceable to their physical location

Licensed hardware systems would come equipped with (or, in the case of pre-pause hardware, be retrofitted for) on-chip location attestation (e.g., latency-based location) to ensure that they remained at their registered location and visible to treaty...

more AI Verified source (Oct 10, 2026)