Catalytic Nuclear War in the age of AI and autonomy: an emerging escalation risk

It no first use, even backed by treaties, is just the prisoners dilemma all over again. Unless you can somehow reliably enforce a prohibition on the development of first strike capabilities, including the development of these in secret, you’ll never be sure that this can’t be used on you and your back to classical Mutually Assured Destruction style escalation.

Also… the biggest issue with such a no first use treaty is that the current champions of “assured retaliation”/“second strike” capabilities are ballistic missile submarines. Which are unfortunately for such a no first use treaty also one of the best first strike weapons platforms thanks to being able to both hold position offshore as close as comfortable and use depressed trajectories extremely quickly.

And with smaller autonomous platforms being developed and miniaturised warheads giving you nukes that fit on “regular size” cruise missiles and the latest generation of submarine launched cruise missiles, you effectively have a situation where you can from a military strategy standpoint be sure that you aren’t vulnerable to such close range first strike capability since just as we started to get the ability to reliably monitor for near surface submarine traffic with magnetic sensing, we also began building a new generation of smaller autonomous vessels that are too small to be detected this way, which could in future even be made from non magnetic materials now that crew safety is no longer a requirement.