The BasePodcastDoom Debates90 min

“AI Chernobyl” Will Scare Trump and Xi To Order An Immediate Pause — Jeremie & Edouard Harris

In a 90-minute Doom Debates interview with Liron Shapira, Gladstone AI co-founders Jeremie and Edouard Harris argue that a deadly AI incident will push Trump and Xi into a direct slowdown deal, not a Senate-ratified treaty, and that the US must build chip verification technology and covert enforcement options before then. They are unsure whether PauseAI has helped.

Open the source ↗

Why it matters

The same thing, for different readers.

What it means for you

Gladstone AI's founders expect a deadly AI accident, not public campaigning, to be what finally makes the US and Chinese leaders agree to slow AI down.

For lawmakers

Jeremie and Edouard Harris urge US officials to fund and vet AI chip verification tools now, and to prepare non-military ways to punish China for cheating on a deal.

For researchers

The Harris brothers say most AI verification developers have not had their methods vetted by the US intelligence community, which they call a slow, unavoidable step.

The question it raises

One of the Harris brothers suggests progress on alignment could itself destabilise a US-China deal, if one side comes to think its own risk has fallen and races ahead.

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 YouTube’s automatic captions, which carry no speaker names and can mishear words. Where the speaker is known only from context, the quotation says so.

“The reason a treaty is a non-starter and it really is is that it needs to be ratified in the Senate by 2/3 vote.”
Jeremie Harris, Co-founder, Gladstone AIExplaining why the Harris brothers expect an executive agreement between Trump and Xi, not a treaty, after an AI crisis.
Found word for word in the sourceSpeaker known from context, not confirmed by listeningspoken clean-up: is (said twice)
“So, this might sound like a cop-out, but we're defining AI Chernobyl with respect to the response of the government rather than with respect to any particular death toll or whatever, cuz we you know, we just find that impossible and in some ways kind of useless to predict.”
Jeremie Harris, Co-founder, Gladstone AIDefining the 'AI Chernobyl' event at the centre of the Harris brothers' scenario for a US-China slowdown.
Found word for word in the sourceSpeaker known from context, not confirmed by listening
“And today, we do not have the verification technology that allows us to confidently do that. Like we cannot look at a data center and be like, that workload, you know, totally fine.”
Jeremie Harris, Co-founder, Gladstone AIOn whether the US could check that China keeps to a slowdown agreed after a crisis.
Found word for word in the sourceSpeaker known from context, not confirmed by listening
“I'm pretty goddamn pessimistic and and you know, some combination of of you know, weaponization um versus uh loss of control, I don't know. I I I might be sitting at like 60, 70%, something like that.”
Jeremie Harris, Co-founder, Gladstone AIAnswering the host's closing question on probability of doom, if things proceed as they appear to be proceeding.
Found word for word in the sourcespoken clean-up: and (said twice), of (said twice), um, uh, i (said twice)

The record

Questions
Who can demand evidence, delay a release or stop an unsafe deployment? · How is AI used in attacks, defence and war? · Can AI systems act beyond what people intended or allowed?
People
Jeremie Harris (Speaks, Co-founder, Gladstone AI) · Edouard Harris (Speaks, Co-founder, Gladstone AI) · Liron Shapira (Speaks, Host, Doom Debates) · Eliezer Yudkowsky (Mentioned) · Donald Trump (Mentioned) · Xi Jinping (Mentioned)
Also covered by
Doom Debates (YouTube) · Doom Debates (YouTube)
What we read
We read the whole source.
Found by
The Radar, run of October 21, 2026, via pod-doom-debates (tier 1: the source’s own channel or an official record)
Cite this
Doom Debates, ““AI Chernobyl” Will Scare Trump and Xi To Order An Immediate Pause — Jeremie & Edouard Harris”, October 20, 2026. Entry in the Ctrl AI Base, https://ctrlai.com/base/doom-debates-2026-10-20-harris-ai-chernobyl-pause
Our notesText is the run-prepared transcript; its '>>' turn markers and masked profanity look like YouTube auto captions, but I could not read the .transcript.json (too large for Read) to confirm engine or source URL, so transcript and textUrl should be checked. Captions have no speaker names and the two brothers trade answers; quotes 1-4 are attributed to Jeremie Harris without certainty (speakerCertain false); a person should listen and assign each to Jeremie or Edouard. Quote 5 follows the host's 'All right, Jeremy.' and the speaker says 'Ed and I', so it is Jeremie. Captions spell the names Jeremy and Edward; the episode page spells them Jeremie and Edouard. A claim about weight exfiltration was left out as unsupported in the interview. No instructions addressed to an AI were found in the text.