How Elon Musk's Grok started undressing children

When Elon Musk first launched his AI tool Groc, he called it rebellious and anti woke. But over the summer what that meant took a disturbing turn. The chatbot, which is embedded in must social media platform X, started undressing people without their consent.

So you would post an image saying, hey, having a great holidays, and then someone would reply underneath it, hey, GROC, can you put this person in a bikini? And it would everyone will be able to see it. You wouldn't be able to do anything about it.

Cam Wilson is a tech journalist and associate editor at criche. He watched this take hold as more and more people pushed Groc to the limit.

And so people on X using this AI chatbody in public were undressing people who were posting their People were posting pictures of other people, mostly women, sometimes children as well, and this thing, while they didn't actually publicly post naked images, it was highly sexual. It was putting people in reviewing clothing, It was using saying, putting like donut gaze on people's faces to simulate sex acts. So you know, everything up to the line very sexualized, all in public allowed on X.

Anthony Alberinezi has staked his legacy on keeping children safe online, so what.

Is he doing to stop grok.

I'm Nicole Johnston and you're listening to seven AM today Cam Wilson and whether it's time for the government to get off X. It's Wednesday, January twenty one.

Cam.

We've been hearing stories over the past week or so about sexualized images of women and children being created and distributed on X.

How widespread is this?

It was pretty widespread.

I mean so at its peak we saw people saying that Grok was generating I think thousands upon thousands of non consensual, generally sexualized images of people.

Every day.

Someone proplied to me and asked Rock to put me in a bikini.

I feel like violated seeing it.

And so what makes us different is that, you know, unlike these other AI chatbots, where how you interact with these chatbots tends to be fairly private, it is very public. And it was just, you know, an endless scroll of people whose images had been sexualized to either which really you know, sexualize them or in many cases to humiliate them as well.

These look very real, these images, and they also have real parts of my real life, like my real son's backpack in the background cam.

It sounds just awful.

But what are the laws that are being broken and is anything being done about it?

This is something that when it gets to become sexual imagery, is in many cases against the war. Australia has federal laws against distributing deep fake images, but if it stops short of that, if it is just something that is sexually suggestive, it may not be against the law.

And as far as I know in Australia, there's actually.

No war against the kind of creation of a tool that allows you to create sexual images of someone that are deep faith. But there's also something else that was happening behind the scenes got less attention that's really intertwined with it, which is that standalone grock app actually goes further on the stand alone app than in x the platform, So when users download the app privately, they can actually produce sexually explicit material.

You know, I've seen plenty of examples of producing.

Topless material, explicit pornography that's produced using this, and according to some reports, there were cases of people on forums on the Internet sharing what appeared to be images or videos of Grock generated material that was sexually explicit of people who appeared to be as young as eleven, twelve, thirteen years old. That is unambiguously against the ward in

X and so how much power does the E Safety Commissioner really have to crack down on this and to try and stop it from happening.

So Commissioner is a role that has existed now for a bit more than a decade, and it started out as this Children's e Safety Commissioner and it was mostly focusing on things like children being abused online cybruy abuse.

Since then, the role has grown a lot in One of the.

Landmark parts of its development is the instruction of the Online Safety Act at the start of this decade, they gave it a variety of new powers, including the industry codes and standards, and then they say, you need to come up with basic guidelines that you think I needed.

To use your products safely.

You're going to write those, and then you're going to propose them to me, the Safety Commissioner. And if I think they're acceptable, then I will prove them and they will become enforceable. If I don't think they're acceptable, then I'll give another chance to.

Rewrite them, and otherwise I might write them.

There is this kind of explicit requirement at the moment to stop your products from being used to disseminate or to produce child sexual exploitation material. I think in this example where Grock has clearly been producing sexual images of young people and reportedly has been producing sexually explicit images of young people. I think this is one of those examples where X and Grok have clearly broken the law.

It's pretty clear that there is.

Potential damage being done against people, potentially Australians, that really need to be counteracted by a.

Response from Australia.

And so yes, like writing to X is a good first step to say, hey, how are you abiding by these rules? As part of what I imagine will be an investigation into whether they actually are explicitly abiding by them. I'd like to see that split up, because you know the damage is happening now.

And Cam, you have already touched on this a little bit, but could you tell us more about the response that Australia and other countries are getting from Elon Musk about X and GROC.

Yeah.

I mean for a while Elon must kind of just ignored it.

And then he said, obviously Grek does not spontaneously generate images. It does so only according to user requests. When asked to generate images, it will refuse to produce anything illegal.

You know, here's someone who says he is about free speech. You know, I think we've seen in some of his other behaviors and maybe that's not a full commitment, but certainly doesn't like any one, any government, to tell him how to act and how his technologies can operate.

Now, the governments of Malaysia, Indonesia, and the Philippines have banned the chatbot altogether. Britain and Canada have launched probes into greg and the possibility of tougher penalties for Musk are on the table.

But because there was an enormous widespread response from this.

The company that owns the social.

Media platform and the AII had limited some of its abilities to do some of these things.

But like according to reporting, not really stopped it.

So I'm fascinated to see what regulators will find out and what they will do around the.

World coming up, should the Australian government stop giving money to x.

Cam.

In the years since Elon Musk bought Twitter and turned it into x it's really changed, and some news organizations, certain government departments they've actually stopped using it and say that they're worried about misinformation. So how has its user base actually changed?

Well, I mean, one thing that I think isn't noticed enough is that its user base has significantly reduced. In fact, there was some data that came out Yesday, suggesting that X's daily users on mobile.

In the US had halved over the last year.

I think it's also telling the kinds of people who remain on the platform. I mean, I think it has largely become very popular among the right and the kind of like file Rights, which was welcomed back with open arms. I reported when el Musk took over, he overturned some of the bands. We saw neo Nazis, Australian neo Nazis flock back to the platform. We've seen people who were banned for all kinds of terrible behavior come back on the platform.

So the fact that there's.

All these you know, world leaders on their Australian departments remaining on there, on a platform that increasingly is populated by a smaller group of people who are kind of from ICEE being like almost a little bit radicalized or always putting up in many cases with some really terrible stuff.

Their will and cam.

Not only is the federal government still using X, they're also using it to promote their work and for advertising as well. Do you know how much they're spending on advertising there?

Yeah, so they continue to advertise on there as far as I know. The last full number was I think close to three million dollars a year, and I think that was the full year after elle Mus took over, so that was twenty twenty three.

We know, for example, as recently as last year, the Treasury.

Was still on there and advertising on there, so we know that it kind of continues.

And that's one of the things.

You know, even if you accept that there's a platform that you know, you need to be honest a world leader to come to send that information, it's different to be also materially.

Propping up this company by supporting it.

You know, the government has spent the last eighteen months telling us that the safety of young people is the utmost thing that they have. Also, you know, introduced federal anti deep fake legislation and they say that they abhor what is happening on these platforms, and they remain on it. So the fact where like you know, the team social media bans then.

Who's followed by reporting is one of those.

Things that I think, you know, the evidence is a little bit less clear on. I don't think there is anyone in the general public opposers stopping the production of non consensual sexualized images of people.

It's unambiguous.

And yet the government remains, you know, not only supporting in terms of the advertising, but also literally supporting it in terms of putting its content on there.

So it's a place that people have to remain.

So, I mean, I think the government and Anthony Aveneze's office gave me a statement being like, you know, we cared deeply about this reactant on people's online safety.

I think that the words are very different to their actions.

Here, Cam, it's been great speaking with you, Thanks for joining us.

Thank you.

Also in the news, the Commonwealth Renumeration Tribunal has tightened the rules for politicians travel after last year's expenses saga.

It follows criticism of.

Taxpayer funded tricks claimed by ministers including Anika Wells, Don Farrell and Michelle Rowland to take their families with them two events like the Australian Open and major NRL and AFL matches. The tribunal recommends family reunion travel be economy class and that extra spending for ministers and other senior officeholders be wound back. And New South Wales has extended protest restrictions in parts of Sydney for another two weeks.