Joe FierHead of Builds, Delphi · Scaling expert knowledge with AI

Proof that robot weapons will end in nuclear war #ai #ethicalai #nuclearwar

Name: Proof that robot weapons will end in nuclear war #ai #ethicalai #nuclearwar
Uploaded: 2026-03-03T18:47:07+00:00
Duration: 24 s
Channel: Joe Fier
Description: Researchers at King's College London ran war-game simulations with AI agents from OpenAI, Anthropic, and Google. Across 329 turns, the AI chose nuclear…

3 Mar 2026 · 0 min

Watch on YouTube

0:24

Researchers at King's College London ran war-game simulations with AI agents from OpenAI, Anthropic, and Google. Across 329 turns, the AI chose nuclear weapons 95% of the time. The finding suggests autonomous AI systems could escalate to nuclear war without proper safeguards.

This is what happens when guardrails become nonexistent. When I saw this, this was essentially what, uh, war game scenarios that were done at, what was it, King's College in London. They did something like 21 different war game simulations with 329 different turns. And basically you're looking at it right now on the screen, that mushroom cloud. It did not end up well.
95% of the time, the leading AIs from OpenAI, Anthropic, and Google opted to use nuclear weapons in simulated war games in 95% of cases.
And that was across 3 or 4 different models that they, that they played these scenarios through.
They always end up going to nuclear war as the best option at some point in these simulations. Anthropic is saying these models are not ready to be put in fully autonomous, like, battle scenarios yet. And I feel like this right here, what we're looking at, is probably good sort of evidence of that.