I have long supported old-style platform immunity protecting providers from liability for the content of a search result, forum post, or the like.

Both individually tailored content recommendation algorithms and LLM output are different, and companies should be liable for the output of both when they're used in the way social media platforms and search engines are currently using them.

Governments would love a data center to be suddenly filled with digital persons that haven't paid taxes so far. But if a companion deletes aan AI, is that murder? The whole thing gets silly really fast.

I really hope more courts pursue this path. If a car was put on market that was found to have a manufacturing defect that's caused nearly 40 deaths, it'd be taken off market right fast. If a food product is sold that's found to have been contaminated, it doesn't even need to have caused harm for it to be pulled off the shelves.

It's absolutely bananas that we have a technology that the creators themselves say they've lost control over on multiple occasions, that's been proved to exacerbate problems for people with mental illnesses, and has so far played a part in several dozen suicides and homicides including at least two mass shootings, just be freely accessible and encouraged to be adopted left and right by absolutely everyone.

Exactly, if they’re presenting false information in response to a query, then the machinery that generated that false information is an implementation detail. Shouldn’t matter who or what wrote it. If Google’s logo is right there, then they are the ones responsible for it.

Stupid on its face. The whole compromise was that search engines were just indexes and so not necissarily responsible for what they link users too (within limits), and websites encourage their bot because it sends them traffic.

Then they blow that all up with a "summary" no one asked for.

I think this is more because to them, Google is a foreign entity requiring more regulation just like how the US regulates TikTok and how China regulates US companies.

Yeah, if by "rouge" you mean purposely taking guide rails off a model, giving it access to hacking tools, instructing it to hack, auto-approving everything the LLM does, and letting it run with no human oversight for long periods of time

There is plenty of opportunity for OpenAI, Anthropic, et. al to take a step back and think about what they're doing before dropping the brick on the accelerator. IDK how they're not on the hook for any of this.

“rogue”

Yeah, "rogue"

court case not legislation

It's who got sued, but I assume that it sets legal precedent for all. I'm not German though, so I don't know how their legal system works.

Because you start with one to set the example then spin out from there.

It's very rare and very hard to just blanket regulate something up front before you have examples of failure modes into the problems.

It kind of ties into the whole concept of innocent until proven guilty.

Nothing is allowed to ever create a problem. Then you stifle innovation change and improvement way too hard.

The problem is the lag time from the new thing to the regulation being put into place and all the damage that can be done in that lag time.

It's why the saying the law lags behind new technologies is a thing.

It's one of the few things we have to just accept if we want to have progression. We are totally allowed to though b**** moan wine and cry that it takes way too long for the reaction to happen cuz it does.

It's an excusable for how long that lag time is. Some lag time is unavoidable, but it does not need to be as bad as it is.

In the US maybe...