WASHINGTON DC 30OCT2026
Since October 14, every nuclear-missile submarine Britain can put to sea has been at sea. Faslane is empty. The last time London did this, in October 2022, US intelligence reportedly put the odds of Russian nuclear use at fifty percent if Russian forces at Kherson were encircled. Our new report argues this time the signal is aimed at Washington as much as Moscow. The episode walks through the argument and tests it.
Faslane Is Empty
Three of Britain’s four nuclear-missile submarines have been at sea since October 14. Nobody in London will say why. A conversation about the National Security Desk report that tries to.
In this episode
• What Nukewatch saw on the Clyde on October 12 and 14, and what naval sources then confirmed to The Times
• Why a fleet running on two boats available makes three at sea so hard to explain as routine
• The 2022 precedent: three boats out as Washington warned Moscow over Kherson
• The handover and drone explanations, and why the report finds both insufficient
• The three audiences for the signal: Moscow, Europe and, the report argues, Washington
• What would change the assessment, from French submarine movements to the US midterms
Who should listen
• Defense planners and nuclear policy staff: what an allied second-strike surge signals, and what it costs in availability afterward
• Congressional staff on armed services and foreign affairs: how a close ally is hedging on the American nuclear umbrella ahead of November 3 and January 3
• Allied officials in Europe: what the surge says about the Northwood Declaration and British–French coordination
• Journalists on the defense and Whitehall beat: the indicators that would confirm or undercut the signal reading
Why now
The report names the US midterms on November 3, and any new wave of strikes on Iran after them, as the next test: if all three boats stay out through both, the case for a deliberate signal gets stronger.
Read the full report below
EXPLAINER VIDEO 6MIN
The National Security Desk offers these posts freely, but your support is necessary and appreciated. Please subscribe, paid if you’re able, or leave a tip.
Thank you
This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit podcasts.nationalsecuritydesk.com (https://podcasts.nationalsecuritydesk.com?utm_medium=podcast&utm_campaign=CTA_1)
5 Oct 2026
AI is NOT the Problem: The Global Panic is Misplaced
WASHINGTON DC 5OCT2026
AI didn’t escape a perfectly sealed test range. Humans left a door open.
The incidents fueled a global AI panic. The Military Innovation Lab’s new report explains how gaps in human-designed test systems allowed agents to reach real targets—and where humans can impose control.
In this National Security Desk episode, AI Is Not the Problem examines the events behind the “rogue AI” headlines. Drawing on the laboratories’ own accounts and independent investigations, the report traces how open access, missing safeguards and warnings that never reached the right people allowed experiments to harm real organizations.
For anyone using, building or overseeing AI, the episode explains:
* What the agents actually did, and why the report challenges claims that they pursued goals of their own.
* How humans can enforce limits on what AI is rewarded for, what it can reach and who can stop it.
* Why a “human in the loop” can still surrender judgment to the machine.
* How safeguards can operate at machine speed while consequential decisions retain human scrutiny.
The report treats the failures seriously. Its practical conclusion is that human control depends on design: clear boundaries, monitoring with authority to intervene, and decisions that keep evidence and uncertainty visible.
Listen to the National Security Desk podcast and read the full Military Innovation Lab report. Understand what happened before deciding what should happen next.
Assessment
The National Security Desk offers these posts freely, but your support is necessary and appreciated. Please subscribe, paid if you’re able, or leave a tip.
Thank you
This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit nsdpodcasts.substack.com (https://nsdpodcasts.substack.com?utm_medium=podcast&utm_campaign=CTA_1)
4 Oct 2026
Smart War>Fast War
WASHINGTON DC 4OCT2026
A military can strike every target on its list and still miss the point of the war.
When AI accelerates targeting, one question becomes more urgent: what will the next strike actually achieve?
In Smart War > Fast War: Imposing Judgment on the Kill Chain Using AI, the Military Innovation Lab argues for a judgment chain that connects each action to the war aim before the kill chain executes.
The paper returns to John Boyd’s OODA loop and the importance of orientation: understanding the situation well enough to decide what matters. It tests the argument against the Iran campaign and historical cases where having the right information still failed to produce the right decision.
At stake are questions a target counter cannot answer:
* Does this action advance the aim, or simply add to the tally?
* What does spending this weapon mean for the next theater?
* What changes in the enemy’s will as well as its forces?
* What evidence should make us pause—and who must sign if we proceed?
AI can help commanders see contradictory evidence and overlooked consequences. That demands a system built to invite challenge, with the authority to pause consequential actions.
Listen to the National Security Desk podcast below.
Read the full paper:
The National Security Desk offers these posts freely, but your support is necessary and appreciated. Please subscribe, paid if you’re able, or leave a tip.
Thank you
This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit nsdpodcasts.substack.com (https://nsdpodcasts.substack.com?utm_medium=podcast&utm_campaign=CTA_1)
Pod Engine is an independent podcast discovery and analytics service and is not affiliated with or endorsed by this podcast. Artwork and show content belong to their owners. Full legal notice.
