Slate Business What Next: TBD | If You Give A.I. a Nuke
Nov 30, 2025
Josh Keating, a senior correspondent at Vox and expert on foreign policy and nuclear security, dives into the complex blend of AI and nuclear systems. He explains the outdated nature of the current nuclear command, contrasted with the alarming push for integration of AI in military strategies. Topics include the potential for AI mishaps, the risks of an AI arms race with China, and the precarious relationship between military leaders and tech companies. Keating stresses that while AI can enhance operations, it poses significant risks that must be carefully managed.
AI Snips
Chapters
Transcript
Episode notes
Brzezinski's False Alarm
- In 1979 Zbigniew Brzezinski nearly woke President Carter to report 220 missiles detected off Oregon.
- A faulty computer chip later proved the alarm false, averting a disastrous call to launch.
Petrov's Moment Of Restraint
- In the 1980s Soviet officer Stanislav Petrov judged a reported U.S. missile launch to be a sensor false positive.
- His decision to wait confirmed the alarm was sunlight on clouds, preventing possible retaliation.
Low-Tech Systems, High-Stakes Decisions
- Much of the nuclear command-and-control system remains low-tech and uses human judgment at key moments.
- Introducing AI risks blurring where humans must decide and where machines inform, complicating accountability.