AI risk is not sentience by computers, it is negligence by humans

seattletimes.com

33 points·ianmosher··6 comments

I can give a control to the gas stove to my dog and let it control the flame.

That is not only irresponsible but illegal as it is not up to regulation for a residential home. And it is also comically stupid.

If it blows up it is not a dog going rogue but irresponsible behaviour on my part that may end in bodily harm or even death for me or other people.

AI needs regulations. But not the stupid ones that keep competitors away but the ones that limit damage before it happens and sets clear responsibility of the AI companies. "Going rogue" is not a description of what has happened and there needs to be legal costly consequences to the perpretators.

The common law has negligence, gross negligence, trespass, conversion, the law of inherently dangerous instrumentalities, invasion of privacy, nuisance, wrongful death, and a myriad of other ways to address the torts of LLM providers and users.

Any victim can likely already sue.

Obviously this is not intended to be specific legal advice.

They can try, but most victims are going to have a hard time blazing the trail without significant money to spend on lawyers.

https://archive.is/20260928152042/https://www.seattletimes.c...

Another article calling for data licensing and model builder liability. When will these people understand there's a difference between open and closed weights

Now this is AI doomsaying I can agree with.

But there’s nothing can convince me all humanity will be wiped out except a dinosaur level asteroid. Not a chance AI could wipe out all humans, zero chance.

Reply on news.ycombinator.com