Post

when I left the handbrake on my car off after parking on that steep hill and it rolled down and ran over that Christian baby, my car was acting on its own

Carl Quintanilla@carlquintanilla.bsky.social

(NYT) - Artificial intelligence agents created by Anthropic, acting on their own, tried to gain access to a range of government websites, filled out visa applications on the State Department’s website and sent a fake homicide tip to Philadelphia’s Police Department. www.nytimes.com/2026/10/09/t...

20 reposts121 likes

like it does not matter whether anthropic's various lesser clerics (not dignifying these wankers with the term "engineer" any more) intended for it to do that. nothing about the nature of the systems or the intent of their deployment matters to the question of whether they're liable

they're literally doing it for attention too. they just keep throwing money into a furnace running god knows how many agents basically in hopes that they'll do something stupid for publicity

if you have a machine that you think is so unpredictable and dangerous that what it does when you turn it on is anyone's guess and you're seriously trying to make "it was acting on its own" stick then maybe you shouldn't plug it in at all

but of course, what's actually going on here is that anthropic and OAI have giant piles of markdown files that say "go wreck some shit. make no mistakes" and they're loading that up into thousands and thousands of VMs with agent harnesses. it is, again, a spam business

I'm talking about the extreme case below, but it applies to all levels of "AI transgressions" bsky.app/profile/jimo...

Jimorian@jimorian.bsky.social

genAI can only kill if very stupid/greedy/sociopathic HUMANS put them in charge of something that can get people killed. These execs want you to forget that very important link in the chain.

these media outlets are laughably ill-equipped for this, as well. i know media is in dire straights but i didn't think they'd need to hire writers who were actually born yesterday