> If you produce a product that causes harm, and there are steps that could be taken to prevent that harm, you should be held responsible for it.

This is wholly unrealistic. Any product can be used to cause harm and there are always steps that could be taken to prevent that. E.g. ceasing sales. But that would often do more harm than it prevents.

I appreciate the pushback. I’m reading your argument that every product can be used to cause harm, I agree with that take. The question is did the manufacturer do everything reasonable to limit the harm caused?

You can’t go after a company that makes kitchen knives if those are used to harm bc there’s nothing reasonable they could have done to prevent that harm, and there’s a legitimate use case for knives for cooking.

In this case, my understanding is other companies (OpenAI and Anthropic) have done more to limit harm, whereas XAI hasn’t.

Personally I can't help but think that 'reasonable' is a dangerous legal standard due to its unpredictability, subjectivity and assumed values and knowledge. Is it reasonable to put powdered aluminum and iron oxide into paint? What about when the paint is going onto a Zeppelin? Oh wait, those are thermite's ingredients. Oops. Is it reasonable for the paint seller to be held liable for selling paint with common reagents?

Things aren’t black and white, and that’s why we have humans and the law. There’s no clear definition of what probable cause means in search warrants but it’s “subjectivity” does not mean you should have no searches?

But in this case it’s pretty easy, other model providers have in fact limited harm better than grok. So you don’t even need something arbitrary, just do it as well as competitors.

> The question is did the manufacturer do everything reasonable to limit the harm caused?

OK. A different and better question.

The problem is, would it be considered reasonable to avoid harm to the mental wellbeing of bikinified persons at the cost of harm to all users enjoying a service supported by bikinification earnings.

I can use Photoshop to create a sexualized image of someone irl.

How's that any different?

If you were to provide Photoshop as a Service on a sufficiently large scale, you would also be expected to take all reasonable measures to prevent it being used to disseminate CSAM and other abusive material.

So, no different to the standard that X should be held to.

And people do do this and have been making crazy and creepy pictures online since its inception. It's never been that much of an issue until now.

In UK law, it isn't.

The practical difference is simply that now it is happening far more frequently.

I guess the difference is that it requires a certain amount of time, ability and skill to make creepy pictures using photoshop - which limits the number of people who will actually do that.

Photoshops also does not have a direct distribution channel built in, which means that if even if you had the wherewithal, knowledge, and nothing better to do, your creepy images would likely stay on your computer, and never see the light of day.

As I understand it, with Grok you simply give it a sexualised prompt and it does it for you in seconds, and immediately distributes the results to a potential audience of hundreds, thousands or even millions of people, where it will likely stay for a long period of time.

To my mind that's definitely rather different.