The BasePodcast80,000 Hours109 min
Tom Milton on how we could verify a US–China deal on AI
In a 109-minute 80,000 Hours interview, Amodo Design CEO Tom Milton describes a pilot that reruns a cluster's inference on a separate verifier and checks its network traffic, argues that such verification would let a US–China training pause continue without shutting data centres, and estimates that about 50 engineers worldwide work on compute verification.
Why it matters
The same thing, for different readers.
What it means for you
Tom Milton says that without checking tools, a US–China pause on AI training would likely mean simply switching data centres off.
For lawmakers
Tom Milton urges governments and AI labs to red-team compute verification tools now, because hardware such as network certifiers has long lead times.
For researchers
Amodo Design reports a pilot on about 32 GPUs that checks a cluster serves only an agreed model; zero-knowledge versions are still in development.
The question it raises
Tom Milton raises whether nations can trust each other on AI without seeing each other's models, drawing on nuclear and chemical weapons inspections.
In their own words
Every quotation was found, word for word, in the text we saved from the source. That shows the words are there; it does not make them true, or prove the speaker meant them as quoted.
From the publisher’s own transcript.
“Zero. I think we would probably just shut down the data centres.”
“And lots of what verification is doing is reducing the cost to complying with some kind of slowdown. So at first it might be turning the data centre on and off, then it might be serving only inference”
“So maybe only 50 engineers in the world are working on AI verification right now, on compute verification.”
“I think the current position is something like: it’s tractable but not solved. There are small but competent teams working on this, but we still need a moonshot R&D project to happen.”
“The thing that would most help me sleep at night is if the major AI developers had dozen-person teams who are focused on making sure that verifiable deals can happen.”
The record
- Questions
- Who can demand evidence, delay a release or stop an unsafe deployment? · Are AI companies putting stronger safeguards into practice?
- People
- Tom Milton (Speaks, CEO, Amodo Design) · Tom Reed (Speaks)
- Also covered by
- 80,000 Hours (YouTube)
- What we read
- We read the whole source.
- Found by
- The Radar, run of October 24, 2026, via
pod-80-000-hours-podcast(tier 1: the source’s own channel or an official record) - Cite this
- 80,000 Hours, “Tom Milton on how we could verify a US–China deal on AI”, October 23, 2026. Entry in the Ctrl AI Base, https://ctrlai.com/base/80000hours-2026-10-23-tom-milton-ai-verification