At least half a dozen nations are working on such systems now. They're doing this not because they think it's a good idea, but because there's this attractor basin they recognize we reach by default, of a new arms race of faster, smarter, stronger AI weapons spiraling up. This is something we don't think is a good idea.
This letter is essentially a petition, which we'd like to take to the U.N., showing that the AI and ML communities don't want their work used in autonomous weapons, things that are built specifically to, by themselves, offensively target and kill people. Having this sort of grass roots effort has precedent: chemical weapons, landmines, and laser-blinding weapons were all banned globally with bans and treaties based on this sort of thing. It's true that terrorists and rogue states might still use them in isolation, but you won't get this effect of the major powers having an arms race. Although these things are under development right now in multiple countries, they haven't been deployed to the field yet. So we're really at an inflection point: we're trying to get a ban in place before they're actually deployed at all, because after that it'd be much harder to get such a treaty. If you agree with these sentiments, we are collecting signatories from the community.