Olivia Solon

These are very serious allegations.

Clearly, we're dealing with child sexual exploitation material, which is illegal for platforms to host and distribute.

And then there's non-consensual intimate imagery, which is colloquially known as revenge porn.

And even these types of images, if they are generated by AI, in many countries, they're still considered to be illegal to create and to share.

There's enough evidence that XAI is not removing these items as quickly as it should be doing.

It could face pretty serious penalties, perhaps starting with fines and then moving through to even blocking the service in certain countries, as we've already seen.

So Ofcom has a very lengthy process, which they keep on sharing with us, which is a sort of, I think it's like a dozen steps, and we're at steps like two or three at the moment.

But in theory, you know, after it'll have some back and forth with the company, interrogate whether the company has taken action quickly enough to remove this material, and then ultimately it could issue fines and even block the service in the country.

I think the issue here is going to be that

most of the images posted to X were not super explicit.

They're people in underwear or bikinis, and it's being done without people's consent, but they might not meet the bar for illegal material, non-consensual intimate image abuse, or AI-generated CSAM.

There are very specific criteria to meet those thresholds, and I think there has been a little bit of a conflation between

Pictures of people in bikinis that have been created non-consensually, which may potentially meet the bar, but may not, and revenge porn.

And similarly, the same, unfortunately, with pictures of kids, putting a kid in a bikini does not make that child sexual exploitation material, however unpalatable you might find that.

Well, I think there's another aspect we've got to consider, which is the guardrails that have been put into Grok and Elon Musk has been experiencing.

explicit that grok has fewer guardrails it's more permissive than other models thinking about like chat gpt and google's gemini in terms of creating more risque images so that has put it in the firing line yes and the kind of automatic distribution platform it has in the shape of x which is unusual the other models don't typically immediately publish their outputs although i believe sora does have a standalone app which is open ai's video generation model

And then Elon Musk obviously is in the firing line.

He's been this, you know, bit of firebrand talking about free speech, meddling in British politics.

So I'm sure the government is happy to engage in a sparring match with him.

But I don't believe in this case, this is a politically motivated attack on the platform.