The Daily · 2026-09-24
What Safeguarding Nukes Can Teach Us About Safeguarding A.I.
This is Wakewire's analysis, based on the recording and the passages quoted here. Tone and candor labels are our read, not a finding about anyone's motive. Our standards
- Source
- The Daily · Episode page
- Released
- 2026-09-24
- We read it
- 2026-10-02
- Recording
- Recording has gaps
- Read confidence
- low
- Review
- Passed automatic checks
What the episode covered
David Sanger said nuclear nonproliferation is the main available model for AI safeguards, but AI is harder because governments do not fully control it and the technology moves faster. The register is mixed, with institutional framing, concrete limits, and a transcript that cuts off mid-answer. Watch whether U.S.-China talks produce only another narrow sentence or real expert mechanisms.
Our summary, written from the transcript. Not the speakers' words.
The episode, in order
01Commercial opening
It establishes a named sponsor segment separate from the editorial discussion.
02News setup
The episode frames AI safeguards as a live geopolitical question between the United States and China.
03Nuclear precedent
This is the historical model the conversation uses to evaluate possible AI controls.
04Why the analogy is limited
The same control points are weaker for AI, where compute, private companies, and smaller actors matter.
05U.S.-China obstacles
A safety agreement has to compete with strategic and economic incentives.
06Possible next steps
The practical path described is incremental and procedural rather than a full treaty.
Sponsors named in this episode
| Sponsor stated | Where, claim and disclosure |
|---|---|
| YouTube Premium |
|
No sponsor named doesn't mean no ads. Each row was taken from the ad break as it was read on air. The verbatim passage behind every row is checked against the transcript character for character. Short excerpts appear below.
Roles and stated affiliations
Sponsor or conflict flagged. Named commercial segment — The recording includes a YouTube Premium ad, and no editorial relationship between that sponsor and the AI or nuclear discussion is stated on the recording.
Wakewire's read of interests
- Michael Babarro: He hosts and frames the discussion around the Trump-Xi meeting, AI risk, and the nuclear analogy.
- David Sanger: He is presented as a New York Times colleague explaining the historical comparison between nuclear safeguards and AI safeguards.
- Sean Evans: He voices the YouTube Premium promotion in the opening commercial segment.
Recording limits
- The transcript cuts off mid-sentence after Michael Babarro says, "ultimately, everyone did agree to sign that".
Claims that need more support
- David Sanger said heads of frontier companies put the chance of civilizational extinction at 10%, but the episode excerpt does not name the executives or source the figure.
Claims made on air that the episode itself did not back up.
Questions the conversation didn't reach
- The episode excerpt raises but does not settle how the one-sentence U.S.-China understanding on AI agents and nuclear weapons would be defined or verified.
Topics the episode did not address. This does not mean anyone avoided them.
Are you from this show? You can ask for a correction, reply, or ask us to stop coverage on the rights page.
The read
The read is that Sanger is not saying the nuclear model cleanly transfers to AI. He is using it as the only tested starting point, then narrowing the realistic outcome to small confidence-building steps. The charitable read is that nuclear history gives policymakers a proven starting place. The boring read is that the likely output is a committee or another sentence. The skeptical read is that the analogy breaks where AI lacks state control, fixed materials, and stable treaty timelines.
This read assumes: The supplied transcript captures only part of the episode because it ends mid-sentence. The named ad segments are separate from the editorial discussion.
Every opinion labeled as one. This is Wakewire's analysis of the recording, not a finding about anyone's motive.
What was said
Commercial opening
A YouTube Premium ad promotes offline downloads, background play, an ad-free experience, and a two-month free trial.
News setup
Babarro says Trump and Xi are meeting in Washington and may discuss the risks of artificial intelligence.
Nuclear precedent
Sanger says the United States and the Soviet Union faced a terrifying technology with commercial uses and extinction risk, then eventually reached agreements that helped prevent nuclear use in anger after 1945.
Why the analogy is limited
Sanger says nuclear control was simpler because states controlled the technology, fissile material was hard to obtain, and launch authority was limited.
U.S.-China obstacles
Sanger says the United States and China view AI as tied to economic survival and national dominance, while their leaders do not share the same view of the threat.
Possible next steps
Sanger says there is no remote chance of a broad treaty-style agreement this week, but he could imagine a second sentence, an experts committee, early warning, model sharing, and independent inspection mechanisms.
Our paraphrase, not the speakers' words.
The passages we relied on
"Well, this week, as President Xi and President Trump meet, no, I don't think there's even a remote chance of that. I think we would be lucky to get a second sentence with a joint experts committee that will meet…"
"If you like YouTube, you'll love YouTube Premium. Hi, I'm Sean Evans from Hot Ones, and I want to tell you about YouTube Premium. It has offline downloads, so you can watch without Wi-Fi, background play, so you can lock…"
"If you like YouTube, you'll love YouTube Premium. Hi."
Short excerpts, credited to the show, at most 40 words each and 150 per episode. We do not publish transcripts. We link to the episode so you can hear it from the show itself.
What to watch for
- A joint Trump-Xi statement adds only a narrow AI safety sentence or announces an experts committee. (This meeting cycle and any immediate readout.)
- U.S. and Chinese officials define what it means to keep AI agents away from nuclear weapons. (Next official AI safety talks.)
- Frontier AI labs accept outside inspectors with authority to raise public alarms. (Next major model-development or safety-policy announcement.)
Written by Wakewire's system under our editorial standards. Reading is open to everyone. A free account personalizes your brief.