I have been stating over and over again that I am much more afraid of the immediate threat of humans leveraging narrow "AI" to increase their shitty behavior, than I am of some theoretical AGI in the future.

The best/worst example of laundering responsibility has to be Lavender: https://news.ycombinator.com/item?id=39918245

This is the same issue as with self-driving cars. The vendor will never take responsibility so who will?

This is why they forcibly keep the human in the loop in those cars. The same with AI. Every AI has "always double check your answers" stamped over it. Yet the unreliability is downplayed. The messages are just there so they have a finger to point when it goes wrong.

This is exactly the type of thing that's worrying about AI and I'm glad the EU is trying to hold this back a bit.

So what happens if someone makes a robot that has a small chance of murdering any given passerby? No human involved in the specific act, thereby no crime?

LLMs aren't actual artificial intelligence and all they're doing is regurgitating words around, it can say output any sort of random nonense, why would that matter? if someone uses what is being spewed as evidence or copy pastes it into an article then the liable ones are the ones who took it out of context and used possibly hallucinated words as "evidence" of any sort.

On LLMs you have warnings about it: OpenAI: "ChatGPT can make mistakes. Check important info.", on copilot: "Copilot uses AI. Check for mistakes."

This is like opening lord of the rings and sueing the tolkien estate because it claims Arda shaped the world, when it is very clearly work of fiction.

I would really like the FTCs of the world to require warnings on all LLM powered products to be as prominently displayed as they are on cigarettes in the EU.

Not just on the source UIs, but required on the contracts for, and UIs of all derivative products. At the moment there are utterly misinformed people running wild with "AI." Example:

https://www.politico.eu/article/tony-blairs-ai-mania-sweeps-...

Disclaimers aren't magic words that dispel all liability. Joe Rogan isn't allowed to lie about people and claim "hey, I'm a comedian, I wasn't being serious." There is a high bar for satire or rhetorical exaggeration around public figures, but this Copilot confabulation is plain slander and Microsoft deserves to get in trouble for it.

Look at the Horizon postal scandal and how long it took to hold anyone responsible.

There’s no _crime_, presumably, but far less clear that there’s no civil case.