And the AI is at fault for this sentencing, not the school authorities/prosecutors/judges dishing justice? WTF.
How is this an AI problem and not a legal system problem?
You can’t just “undo” some girl being harassed by AI generated nude photos of her, so we…
Yes, we should have some protections or restrictions on what you can do.
You may not understand it, either because you aren’t a parent or maybe just not emotionally equipped to understand how serious this actually can be, but your lack of comprehension does not render it a non-issue.
Having schools play whack-a-mole after the photos are shared around is not a valid strategy. Never mind that schools primarily engage in teaching, not in investigation.
As AI-generated content gets less and less distinguishable from reality, these incidents will have far worse consequences and putting such power in the hands of adolescents who demonstrably don’t have sound judgment (hence why they lack many other rights that adults have) is not something most parents are comfortable with - and I doubt you’ll find many teachers, psychiatrists and so on who would support your approach either.
No, but if you send those people who made and distributed the AI nude of her to jail, these problems will virtually disappear overnight, because going to jail is a hugely effective deterrent for most people.
But if you don't directly prosecute the people doing it, and instead just ban Grok AI, then those people will just use other AI tools, outside of US jurisdiction, to do the same things and the problem persists.
And the issues keeps persisting, because nobody ever goes to jail. Everyone only gets a slap on the wrist, deflects accountability by blaming the AI, so the issue keeps persisting and more people end up getting hurt because those who do the evil are never held directly accountable.
Obviously Grok shouldn't be legally allowed to generate fakes nudes of actual kids, but in case such safeguards can and will be bypassed, that doesn't absolve the humans from being the ones knowingly breaking the law to achieve a nefarious goal.
Youths lack judgment, so they can’t vote, drink, drive, have sex or consent to adults.
A 14-year-old can’t be relied to understand the consequences of making nudes of some girl.
Beyond that, we regulate guns, speed limits and more according to principles like “your right to swing your fist ends at my nose”.
We do that not only because shoving kids into jails is something we want to avoid, but because regulating at the source of the problem is both more feasible AND heads off a lot of tragedy.
And again, you fail to acknowledge the investigative burden you put on society to discover who originated the photo after the fact, and the trauma to the victim.
If none of that computes for you, then I don’t know what to say except I don’t place the right to generate saucy images highly enough to swarm my already overworked police with requests to investigate who generated fake underage porn.
Teenagers do stupid shit all the time. But they still get prosecuted or convicted when they do crimes. They go to juvy or their parents get punished. Being 14 is not a get out of jail free card.
The problem stems from parents lack of parenting, a huge lack of real after-school programs, and the tiktokification of modern society.
30 years ago, we had a lot of the same "slap on the wrist" punishments because it was assumed when you got home your parent was going to beat your ass. That isn't a thing anymore (rightfully), because parenting through threat of violence just leads to those kids becoming violent parents.
Our problem is we never transitioned from violent parenting into any other kind. I watched my nieces and nephews get parented by YouTube and get social media accounts before they were 10. COVID created a society of chronically online children who don't know how to interact offline.
And yes, the tools to create bad shit are more accessible than ever, and I always come off as some angry gate keeper, but so much of the internet as it is today has become too easy to access by people incapable of the critical thinking required for safe use.
In the last 5 years, generative AI has taken over most of the "public facing" internet, and with internet literacy at the same level it was 20-30 years ago, we are back in the "walled garden" AOL era, but it is Facebook, Instagram, Twitter, TikTok that are the gardens.
I'm similarly repulsed by the idea of Grok generating images of kids, but if you draw a nude of an adult woman she's not going to get raped by that existing, and you don't have a right to not be embarrassed. Tough shit, deal with it.
In good faith, a few things - AI generated imagery and Photoshop are not the same. If someone can mail Adobe and a photo of a kid and ask for a modified one and Adobe sent it back, yes Adobe’s offices will be raided. That’s the equivalent here. It’s not a tool. It’s a service. You keep using AI, without taking a moment to give the “intelligence” any thought.
Yes, powerful people are always going to get by, as you say. And the laws & judicial system are for the masses. There is definitely unfairness in it. But that doesn’t change anything here - this is a separate conversation.
If not Grok then someone else will do it - is a defeatist argument that can only mean it can’t be controlled so don’t bother. This point is where you come across as a CSAM defender. Govt’s will/should do whatever they can to make society safe, even if it means playing whack a mole. Arguing that’s “not efficient” is frankly confusing. Judicial system is about fairness and not efficiency.
frankly, I think you understand all of this and maybe got tunnel visioned in your anger at the unfairness of people scapegoating technology for its failings. That’s the last thing I want to point out, raiding an office is taking action against the powerful people who build systems without accountability. They are not going to sit the model down and give a talking to. The intention is to identify the responsible party that allows this to happen.
Yes, let's just jail every kid who makes a mistake, ya know, instead of the enablers who should know better as adults...
except for that one guy, let's put him in the white house
Actually you'll see the opposite happen a lot - after Columbine, the number of school shootings went up [0] for example, because before people didn't consider it an option. Same with serial killers / copycats, and a bunch of other stuff.
Likewise, if it hadn't been in the news, a lot of people wouldn't have known you can / could create nudes of real people with Grok. News reporting on these things is its own kind of unfortunate marketing, and for every X people that are outraged about this, there will be some that are instead inspired and interested.
While a lot of punishments for crimes is indeed a deterrent, it doesn't always work. Also because in this case, it's relatively easy to avoid being found out (unlike school shootings).
[0] https://www.security.org/blog/a-timeline-of-school-shootings...
Did you see the part in the article about "raided in France" and "UK opens fresh investigation"?
You definitely can. You don't have to prosecute and send a million people to jail for making and distributing fake AI nudes, you just have to send a couple, and then the problem virtually goes away.
People underestimate how effective direct personal accountability is when it comes with harsh consequences like jail time. That's how you fix all issues in society and enforce law abiding behavior. You make the cost of the crime greater than the gains from it, then crucify some people in public to set an example for everyone else.
Do people like doing and paying their taxes? No, but they do it anyway. Why is that? Because THEY KNOW that otherwise they go to jail. Obviously the IRS and legal system don't have the capacity to send the whole country to jail if they were to stop paying taxes, but they send enough to jail in order for the majority of the population to not risk it and follow the law.
It's really that simple.
In this case, it's far simpler to prosecute the source.
Everything I said is true.
>Deterrence is known to have a very limited effect on behaviour.
It is insanely effective when actually enforced. It's not effective when the goal is to make it seem ineffective so that people can evade the system.
>In this case, it's far simpler to prosecute the source.
The "source" is a tool that tomorrow can be in Russia or CHina and you can't prosecute.
The certainty or likelihood of being caught if a far more effevtive deterrent, but require effort, focus, and resources by law enforcement.
It's a resource constraint problem and a policy choice. If "they" wanted to set the tone that this type of behavior will not be tolerated, it would require a concerted multi agency surge of investigative and prosecutorial resources. It's been done before, if there's a will there's a way.
I genuinely cannot tell if you are being comically naïve or extremely obtuse here. You need only look at the world around you to see that this does not, and never will, happen.
As another commenter said, this argument is presenting itself as apologia for CSAM and you come across as a defender of the right for a business to create and publish it. I assume you don't actually believe that, but the points you made are compatible.
It is as much the responsibility of a platform for providing the services to create illegal material, and also distributing said illegal material. That it happens to be an AI that generates the imagery is not relevant - X and Grok are still the two services responsible for producing and hosting it. Therefore, the accountability falls on those businesses and its leadership just as much as it does the individual user, because ultimately they are facilitating it.
To compare to other situations: if a paedophile ring is discovered on the dark web, the FBI doesn't just arrest the individuals involved and leave the website open. It takes the entire thing down including those operating it, even if they themselves were simply providing the server and not partaking in the content.
And yet criminals still commit crimes. Obviously jail is not the ultimate deterrent you think it is. Nobody commits crimes with the expectation that they'll get caught, and if you only "crucify some people", then most criminals are going to (rightfully) assume that they'll be one of the lucky ones.
I don't accept this as good faith argumentation nor does HN rules.
For disagreeing on the injection of offtopic hypothetical scenarios as an argument derailing the main topic?
>It should not be possible to create nudes of minors via Grok.
I agree with THIS part, I don't agree with the part where the main blame is on the AI, instead of on the people using it. That's not a bad faith argument, it's just My PoV.
If Grok disappears tomorrow, there will be other AIs from other parts of the world outside of US/EU jurisdiction, that will do the same since the cat is out of the bag and the technical barrier to entry is dropping fast.
Do you keep trying to whack-a-mole the AI tools for this, or the humans actually making and distributing fake nudes of real people?
Other AIs have guardrails. If Musk chooses not to implement them, that's his personal irresponsibility.
Both, obviously. For example, you go after drug distributors and drug producers. Both approaches are effective in different ways, I am not sure why you are having such trouble understanding this.
If the perpetrator is in another country / jurisdiction it is virtually impossible to prosecute let alone sentence.
It is 100% regulatory problem in this case. You just cannot allow this content to be generated and distributed in the public domain by anonymous users. It has nothing to do with free speech but with civility and common understanding of what is morally wrong / right.
Obviously you cannot prevent this in private forums unless it is made illegal which is a completely different problem that requires a very different solution.
The school authorities messed up.
Both are accuntable.
Correction: kids made the pictures. Using Grok as the tool.
If kids were to "git gud" at photoshop and use that to make nudes, would you arrest Adobe?
If kids ask a newspaper vendor for cigarettes and he provides them .. that's a no-no.
If kids ask a newspaper vendor for nudes and he provides them .. that's a no-no.
If kids ask Grok for CSAM and it provides them .. then ?
Meanwhile, the existence/creation CSAM of actual people isn't legal, for anyone no matter the age.
So when Grok provides the illegal pictures then by the same logic it is Grok that is breaking the law.
If parents or school let children roam the internet unsupervised... then?
The explosive sellers that provide explosives to someone without a certification (child or adult) get in trouble (in this part of the world) .. regardless of whether someone gets hurt (although that's an upscale).
If sellers provide ExPo to certified parents and children get access .. that's on the parents.
In that analagy of yours, if grok provided ExPo or CSAM to children .. that's a grok problem,
(Ditto drugs).
It's on the provider to children. ie Grok.
No. That is not how AI nowdays works. Kids told the tool what they want and the tool understood and could have refused like all the other models - but instead it delivered. And it only could do so because it was specifically trained for that.
"If kids were to "git gud" at photoshop "
And what is that supposed to mean?
Adobe makes general purpose tools as far as I know.
Anyone skilled at photoshop can do fake nudes as good or even better than AI, including kids (we used it to make fun fakes of teachers in embarrassing situations back in the mid 00s and distribute them via MSN messenger), so then why is only the AI tool the one to blame for what the users do, but not Photoshop if both tools can be used to do the same thing?
People can now 3D print guns at home, or at least parts that when assembled can make a functioning firearm. Are now 3D printer makers to blame if someone gets killed with a 3D printed gun?
Where do we draw the line at tools in terms of effort required, between when the tool bares the responsibility and not just the human using the tool to do illegal things? This is the answer I'm looking for and I don't think there is an easy one, yet people here are too quick to pin blame based on their emotional responses and subjective biases and word views on the matter and the parties involved.
So let's say there are two ways to do something illegal. The first requires skills from the perpetrator, is tricky to regulate, and is generally speaking not a widespread issue in practice. The second way is a no brainer even for young children to use, is easy to regulate, and is becoming a huge issue in practice. Then it makes sense to regulate only the second.
> People can now 3D print guns at home, or at least parts that when assembled can make a functioning firearm. Are now 3D printer makers to blame if someone gets killed with a 3D printed gun?
Tricky question, but a more accurate comparison would be with a company that runs a service to 3D print guns (= generating the image) and shoot with them in the street (= publishing on X) automatically for you and keeps accepting illegal requests while the competitors have no issue blocking them.
> Where do we draw the line at tools in terms of effort required, between when the tool bares the responsibility and not just the human using the tool to do illegal things?
That's also a tricky question, but generally you don't really need to know precisely where to draw the line. It suffices to know that something is definitely on the wrong side of the line, like X here.
That's actually a good argument. And that's how the UK ending up banning not just guns, but all sorts of swords, machetes and knives, meanwhile the violent crime rates have not dropped.
So maybe dangerous knives are not the problem, but the people using them to kill other people. So then where do we draw the line between lethal weapons and crime correlation. At which cutting/shooting instruments?
Same with software tools, that keep getting more powerful with time lowering the bar to entry for generating nudes of people. Where do we draw the line on which tools are responsible for that instead of the humans using them for it?
The rate of school shootings has dropped from one (before the implementation of recommendations from the Cullen report) to zero (subsequently). Zero in 29 years - success by any measure.
If you choose to look at _other_ types of violent crime, why would banning handguns have any effect?
> Where do we draw the line on which tools are responsible for that instead of the humans using them for it?
You can ban tools which enable bad outcomes without sufficient upside, while also holding the people who use them to account.
The knife (as opposed to sword) example is interesting. In the U.K. you’re not allowed to sell them to children. We recognise that there is individual responsibility at play, and children might not be responsible enough to buy them, given the possible harms. Does this totally solve their use in violent crime? No. But if your alternative is “it’s up to the individuals to be responsible”, well, that clearly doesn’t work, because some people are not responsible. At a certain point, if your job is to reduce harm in the population, you look for where you can have a greater impact than just hoping every individual follows the law, because they clearly don’t. And you try things even if they don’t totally solve the problem.
And indeed, the same problem in software.
As for the violent crime rates in the U.K., I don’t have those stats to hand. But murder is at a 50 year low. And since our post-Dunblane gun laws, we haven’t had any school shootings. Most Britons are happy with that bargain.