Live data from Hacker News

X offices raided in France as UK opens fresh investigation into Grok

bbc.com

621–630 of 1001 posts

Re: X offices raided in France as UK opens fresh investigation into Grok

#621

Earlier quoted context omitted.

This is accountability for the crimes of humans. The crime is creating a system that lets schoolboys create fake nudes of other minors. You don't just get to build a CSAM-generator and then be like "well I never intended for it to be used...". The humans running a company are liable for the product that their company builds, easy as that.

>The crime is creating a system that lets schoolboys create fake nudes of other minors. So like Photoshop? Do you want to raid Adobe's HQ?

Does Photoshop have a "let me jerk off to this minor" button?

Re: X offices raided in France as UK opens fresh investigation into Grok

#622
post #519

Good one. No platform ever should allow CSAM content. And the fact that they didn’t even care and haven’t want to spend money for implementing guardrails or moderation is deeply concerning. This has imho nothing to do with model censorship, but everything with allowing that kind of content on a platform

[flagged]

Re: X offices raided in France as UK opens fresh investigation into Grok

#623
post #568

Earlier quoted context omitted.

You want the French authorities to focus on the Epstein files to the exclusion of all other corporate misbehaviour? Also, it seems pretty likely that Musk is tangled up with the Epstein shit. First Musk claimed he turned down offer to go to the island. Now it turns out Musk repeatedly sought to visit, including wanting to know when the "wildest" party was happening, after Epstein was already known as a child sex abus…

>You want the French authorities to focus on the Epstein files to the exclusion of all other corporate misbehaviour? No i said no such thing, what I said was that the resources of authorities is a finite pie. If most of it goes towards petty stuff like corporate misbehavior that hurts nobody, there won't be enough for the grave crimes like actual child abuser that actually hurt real people. Same how police won't both…

There's no indication that this investigation would draw resources away from investigating the Epstein files. It's happening in France, for starters, whilst the vast majority of Epstein's crimes appear to have happened in US territories. Speaking about "the authorities" as if they're a unified global entity sounds a little conspiratorial.

Re: X offices raided in France as UK opens fresh investigation into Grok

#624
post #519

Good one. No platform ever should allow CSAM content. And the fact that they didn’t even care and haven’t want to spend money for implementing guardrails or moderation is deeply concerning. This has imho nothing to do with model censorship, but everything with allowing that kind of content on a platform

[flagged]

Re: X offices raided in France as UK opens fresh investigation into Grok

#625

[flagged]

Maybe you should look up child pornography laws in Europe. In Sweden, the mere act of scrolling by an image depicting (real or not) a child in a sexual position, and having it stored in the browser cache, is a crime with up to 2 years of prison time.

Re: X offices raided in France as UK opens fresh investigation into Grok

#626

Earlier quoted context omitted.

> This looks like plain political pressure. No lives were saved, and no crime was prevented by harassing local workers. The company made and released a tool with seemingly no guard-rails, which was used en masse to generate deepfakes and child pornography.

No other "AI" companies released tools that could do the same?

In fact, Gemini could bikinify any image just like Grok. Google added guardrails after all the backlash Grok received.

Re: X offices raided in France as UK opens fresh investigation into Grok

#628
post #519

Good one. No platform ever should allow CSAM content. And the fact that they didn’t even care and haven’t want to spend money for implementing guardrails or moderation is deeply concerning. This has imho nothing to do with model censorship, but everything with allowing that kind of content on a platform

[flagged]

The different factors are scale (now "deepfakes" can be automatically produced), and endorsement. It is significant that all these images aren't being posted by random users, they are appearing under the company's @grok handle. Therefore they are speech by X, so it's X that's getting raided.

Re: X offices raided in France as UK opens fresh investigation into Grok

#629

Earlier quoted context omitted.

> This looks like plain political pressure. No lives were saved, and no crime was prevented by harassing local workers. The company made and released a tool with seemingly no guard-rails, which was used en masse to generate deepfakes and child pornography.

> The company made and released a tool with seemingly no guard-rails, which was used en masse to generate deepfakes and child pornography. Do you have any evidence for that? As far as I can tell, this is false. The only thing I saw was Grok changing photos of adults into them wearing bikinis, which is far less bad.

Grok do seem to have tons of useless guardrails. Reportedly you can't prompt it directly. But also reportedly they tend to go for almost nonsensically off-guardrail interpretation of prompts.

Re: X offices raided in France as UK opens fresh investigation into Grok

#630
post #12

Earlier quoted context omitted.

CSAM does not have a universal definition. In Sweden for instance, CSAM is any image of an underage subject (real or realistic digital) designed to evoke a sexual response. If you take a picture of a 14 year old girl (age of consent is 15) and use Grok to give her bikini, or make her topless, then you are most definately producing and possessing CSAM. No abuse of a real minor is needed.

> CSAM does not have a universal definition. Strange that there was no disagreement before "AI", right? Yet now we have a clutch of new "definitions" all of which dilute and weaken the meaning. > In Sweden for instance, CSAM is any image of an underage subject (real or realistic digital) designed to evoke a sexual response. No corroboration found on web. Quite the contrary, in fact: "Sweden does not have a legislativ…

This is the actual law (Brottsbalken 16:10a)

https://www.riksdagen.se/sv/dokument-och-lagar/dokument/sven...

A person who

1. depicts a child in a pornographic image,

2. disseminates, transfers, provides, exhibits, or otherwise makes such an image of a child available to another person,

3. acquires or offers such an image of a child,

4. facilitates contacts between buyers and sellers of such images of children or takes any other similar measure intended to promote trade in such images, or

5. possesses such an image of a child or views such an image to which he or she has gained access

shall be sentenced for a child pornography offense to imprisonment for at most two years.

Then there's Proposition 2009/10:70, which is a clarifying document on how the law should be interpreted:

https://www.riksdagen.se/sv/dokument-och-lagar/dokument/prop...

Let me quote (translated):

"To depict a child in a pornographic image entails the production of such an image of a child. An image can be produced in various ways, e.g., by photographing, filming, or drawing a real child. Through various techniques, more or less artificial images can also be created. For criminal liability, it is not required that the image depicts a real child; images of fictitious children are also covered. New productions can also be created by reproducing or manipulating already existing depictions, for example, by editing film sequences together in a different order or by splicing an image of a child’s head onto an image of another child’s body."

Post reply on HN