Does it give permission to AI companies to create (deep)fakes on demand with no repercussions, though?

What would happen if you generate them at your home with your own model and release them to the public?

When somebody does something, that person is crucified and cancelled (and even prosecuted), but when a company does this, it's called free viral marketing. This is neither fair nor ethical.

What are you upset about?

That a tool is able (and willing) to create an image in someone real's likeness if the user of the tool asks for it?

And you advocate that the tool should refuse?

I personally believe that these tools should refuse to generate erotic images of specific people.

What about non-erotic images?

Imagine you give it a photo of someone (you or someone else) and ask for a fun modification like adding wings or a crown or some cool clothes?

Should that be verboten?

> And you advocate that the tool should refuse?

Yup, and I advocate this on ethical grounds, and I don't care if laws allow this (and even they do not).

> What are you upset about?

Double standards & free pass to corporations. When an individual does or enables this, SWAT will raid their home and take them and their everything in six hours or less.

When a company does this, people go "it's just an image, why so upset".

I believe people shall have dignity and rights. This is why I'm upset.

Why are you so upset that I'm upset about disregard for human dignity?

Thanks, I'm not particularly upset. I was just curious to understand your point of view better.

The reason I'm not super upset is that I believe these things will settle at a reasonable state that societies expect. Especially large corporations will conform to societal expectations.

If you're very worried about deepfakes of actual people, I think you should be more concerned about those models that are NOT the product of a large company, but rather the ones that any random person can run on their computer.

Scale matters. Yes, local models will always be able to run without any guardrails and somebody can use them to create as much objectionable content as they want. But the vast majority of people will interact with these tools through the most convenient mechanism, which isn't a local model. Guardrails can be valuable here even if people can go somewhere else.

And most of those Google images don't even look particularly "spicy" or sexualized. (Except of course for the US American tradition of considering any instance of partial nudity as pornographic, which manifests itself e.g. in YouTube allowing a significant amount of violence while videos with any uncensored breasts are immediately taken down. Which looks rather prudish from a European perspective.)

> Even if people are still outraged, I should also point out that we're trusting what the reporting is saying without any evidence...

The reporters could have easily included the offending images in censored form. The fact that they didn't is somewhat telling.

I'll still point out it's Grok, which does make porn of people already and racist comments and is tuned to be politically incorrect. So there's already enough to be outraged about IMO if you aren't someone like Musk.