Image-generating AI is capable of generating images that are not like anything that was in its training set.
In that case probably the strongest argument is that if it were legal, many people would get off charges of real CSAM because the prosecuter can't prove that it wasn't AI generated.
Better a dozen innocent men go to prison than one guilty man go free?
In this case if they know it's illegal, then they knowingly broke the law? Things are still illegal even if you don't agree with it.
Most (many?) Western countries also ban cartoon underage content, what's the justification for that?
You suggested a situation where "many people would get off charges of real CSAM because the prosecuter can't prove that it wasn't AI generated." That implies that in that situation AI-generated CSAM is legal. If it's not legal then what does it matter if it's AI-generated or not?
That's not quite what I was getting at over the course of the comment thread.
It one scenario, AI material is legal. Those with real CSAM use the defense that it's actually AI and you can't prove otherwise. In this scenario, no innocent men are going to prison, and most guilty men aren't either.
The second scenario we make AI material illegal. Now the ones with real CSAM go to prison, and many people with AI material do too because it's illegal and they broke the law.
This comment thread started with you implying that the AI was trained on illegal material, I'm really not sure how it's got to this point from that one.
Im completely against restrictions on art depictions and writing. Those don't have the dangers of being real but being pawned off as fake.
If it's illegal, and they produce the AI CSAM anyway, they've broken the law and are by definition not Innocent.
To be honest, if it prevents that one guilty man from carrying out such high degrees of abuse to a dozen children, I can't say I'd say no.
I want to stress that this isn't sensationalist grandstanding like wanting to ban rock music or video games or spying on all digital communication in the name of protecting the children. It's just the pragmatic approach towards preventing CSAM in an age where the "know it when I see it" definition of pornographic material is starting to blur the lines.
Well, your philosophy runs counter to the fundamentals of Western justice systems, then.
Why is that? I'd consider this equivalent to the (justified) banning of Nazi imagery in countries like Germany, Austria, Norway, Australia, etc.
No one is harmed by a piece of paper or cloth with a symbol on it, but harm happens because of the symbol's implications.
"Authorized" AI-generated or illustrated depictions of CSAM validate the sexualization of children in general, and should not be permitted, in my opinion. If it enables real CSAM to continue, then AI-generated content is not victimless, and therefore I don't think these hypothetical individuals going to prison for it are necessarily innocent.
It's not the specific thing being made illegal, it's the underlying philosophy of "Better a dozen innocent men go to prison than one guilty man go free" I'm arguing against here. Most western justice systems operate under a principle of requiring guilt to be proven beyond a reasonable doubt, and if there is doubt then guilt cannot be considered proven and the person is not convicted.
The comment I'm responding to is proposing a situation where non-AI-generated images are illegal but AI-generated ones aren't, and that there's no way to tell the difference just by looking at the image itself. In that situation you couldn't convict someone merely based on the existence of the image because it could have been AI-generated. That's fundamental to the "innocent until proven guilty beyond all reasonable doubt" philosophy I'm talking about, to do otherwise would mean that innocent people could very easily be convicted of crimes they didn't do.
I guess we disagree on the criteria for innocent. I don't see possession of such images as an innocent act, especially now that it is impossible to verify what is real or fake.
We aren't disagreeing because that's not what I was addressing in the first place. The comment I'm responding to, from Dave, reads:
In that case probably the strongest argument is that if it were legal, many people would get off charges of real CSAM because the prosecuter can't prove that it wasn't AI generated.
Emphasis added. The premise of the scenario is that possession of such images (ie, AI-generated CSAM) is not illegal. Given that, for purposes of argument, it follows that this would indeed be a valid defense. You'd need to prove in court that the CSAM pictures that you're charging someone with possessing are not AI-generated, in that scenario.
If you want to have a wider discussion of whether AI-generated CSAM images should be illegal, that's a separate matter.
If it has images of construction equipment and houses, it can make images of houses that look like construction equipment. Swap out vocabulary as needed.
Cool, how would it know what a naked young person looks like? Naked adults look significantly different.
It understands young and old.
Is a kid just a 60% reduction by volume of an adult? And these are generative algorithms... nobody really understands how it perceives the world and word relations.
It understands young and old. That means it knows a kid is not just a 60% reduction by volume of an adult.
We know it understands these sorts of things because of the very things this whole kerfuffle is about - it's able to generate images of things that weren't explicitly in its training set.
But it doesn't fully understand young and "naked young person" isn't just a scaled down "naked adult". There are physiological changes that people go through during puberty which is why the "It understands young vs. old" is a clearly vapid and low effort comment. Yours has more meaning behind it so I'd clarify that just being able to have a vague understanding of young and old doesn't mean it can generate CSAM.
But it doesn't fully understand young and "naked young person" isn't just a scaled down "naked adult".
Do you actually know that, or are you just assuming it?
Personally, I'm basing my assertions off of experience with related situations, where I've asked image AIs to generate images of things that I'm quite sure weren't in its training set and that require conceptual understanding to create "hybrids." It's done a decent job of those so I'm assuming that it can figure out this specific situation as well, since most of these models have a lot of examples of naked people and young people in their training sets. But I haven't actually asked any AIs to generate images of naked young people to test this one specific case.
My opinion here is that "naked young person" isn't as simple as other compound concepts because there are physiological changes we go through during puberty that an AI can't reverse engineer. Something like "Italian samurai" involves concepts that occur at a surface level that it can easily understand while "naked young person" involves some components that can't be derived simply from applying "young" to "naked person" or "naked" to "young person".
Well, I haven't gone to any of my image AIs and actually asked them to generate naked pictures of young people. So unless you want to go there this will necessarily involve some degree of theoretical elements.
However, according to the article it's possible to generate this stuff with Stable Diffusion models, and Stable Diffusion models have a negligible amount of CSAM in the training set. So short of actually doing the experiment that would seem to settle it.
I think a lot of people don't appreciate just how surprisingly sophisticated the "world model" that these image AIs have learned is. There was a paper a while back where some researchers were trying to analyze how image generators were working internally, and they discovered that if you were to for example ask one to make a picture of a bicycle it will first come up with a depth map of the image before it starts doing anything to the visual output. That shows that the AI has figured out what the three-dimensional form of a bicycle is based entirely on a pile of two-dimensional training images, with no other clues telling it that the third dimension even exists in the first place.
Just go ask a model to show you, with legal subject matter
What is the AI trained on?
Image-generating AI is capable of generating images that are not like anything that was in its training set.
In that case probably the strongest argument is that if it were legal, many people would get off charges of real CSAM because the prosecuter can't prove that it wasn't AI generated.
Better a dozen innocent men go to prison than one guilty man go free?
In this case if they know it's illegal, then they knowingly broke the law? Things are still illegal even if you don't agree with it.
Most (many?) Western countries also ban cartoon underage content, what's the justification for that?
You suggested a situation where "many people would get off charges of real CSAM because the prosecuter can't prove that it wasn't AI generated." That implies that in that situation AI-generated CSAM is legal. If it's not legal then what does it matter if it's AI-generated or not?
That's not quite what I was getting at over the course of the comment thread.
It one scenario, AI material is legal. Those with real CSAM use the defense that it's actually AI and you can't prove otherwise. In this scenario, no innocent men are going to prison, and most guilty men aren't either.
The second scenario we make AI material illegal. Now the ones with real CSAM go to prison, and many people with AI material do too because it's illegal and they broke the law.
This comment thread started with you implying that the AI was trained on illegal material, I'm really not sure how it's got to this point from that one.
Im completely against restrictions on art depictions and writing. Those don't have the dangers of being real but being pawned off as fake.
If it's illegal, and they produce the AI CSAM anyway, they've broken the law and are by definition not Innocent.
To be honest, if it prevents that one guilty man from carrying out such high degrees of abuse to a dozen children, I can't say I'd say no.
I want to stress that this isn't sensationalist grandstanding like wanting to ban rock music or video games or spying on all digital communication in the name of protecting the children. It's just the pragmatic approach towards preventing CSAM in an age where the "know it when I see it" definition of pornographic material is starting to blur the lines.
Well, your philosophy runs counter to the fundamentals of Western justice systems, then.
Why is that? I'd consider this equivalent to the (justified) banning of Nazi imagery in countries like Germany, Austria, Norway, Australia, etc.
No one is harmed by a piece of paper or cloth with a symbol on it, but harm happens because of the symbol's implications.
"Authorized" AI-generated or illustrated depictions of CSAM validate the sexualization of children in general, and should not be permitted, in my opinion. If it enables real CSAM to continue, then AI-generated content is not victimless, and therefore I don't think these hypothetical individuals going to prison for it are necessarily innocent.
It's not the specific thing being made illegal, it's the underlying philosophy of "Better a dozen innocent men go to prison than one guilty man go free" I'm arguing against here. Most western justice systems operate under a principle of requiring guilt to be proven beyond a reasonable doubt, and if there is doubt then guilt cannot be considered proven and the person is not convicted.
The comment I'm responding to is proposing a situation where non-AI-generated images are illegal but AI-generated ones aren't, and that there's no way to tell the difference just by looking at the image itself. In that situation you couldn't convict someone merely based on the existence of the image because it could have been AI-generated. That's fundamental to the "innocent until proven guilty beyond all reasonable doubt" philosophy I'm talking about, to do otherwise would mean that innocent people could very easily be convicted of crimes they didn't do.
I guess we disagree on the criteria for innocent. I don't see possession of such images as an innocent act, especially now that it is impossible to verify what is real or fake.
We aren't disagreeing because that's not what I was addressing in the first place. The comment I'm responding to, from Dave, reads:
Emphasis added. The premise of the scenario is that possession of such images (ie, AI-generated CSAM) is not illegal. Given that, for purposes of argument, it follows that this would indeed be a valid defense. You'd need to prove in court that the CSAM pictures that you're charging someone with possessing are not AI-generated, in that scenario.
If you want to have a wider discussion of whether AI-generated CSAM images should be illegal, that's a separate matter.
this is the real problem.
AI can compose novel looking things from components it has been trained on - it can't imagine new concepts. If CSAM is being generated it's because it was included in it's training set which is highly suspected as we know the common corpus had CSAM in it: https://cyber.fsi.stanford.edu/news/investigation-finds-ai-image-generation-models-trained-child-abuse
If it has images of construction equipment and houses, it can make images of houses that look like construction equipment. Swap out vocabulary as needed.
Cool, how would it know what a naked young person looks like? Naked adults look significantly different.
It understands young and old.
Is a kid just a 60% reduction by volume of an adult? And these are generative algorithms... nobody really understands how it perceives the world and word relations.
It understands young and old. That means it knows a kid is not just a 60% reduction by volume of an adult.
We know it understands these sorts of things because of the very things this whole kerfuffle is about - it's able to generate images of things that weren't explicitly in its training set.
But it doesn't fully understand young and "naked young person" isn't just a scaled down "naked adult". There are physiological changes that people go through during puberty which is why the "It understands young vs. old" is a clearly vapid and low effort comment. Yours has more meaning behind it so I'd clarify that just being able to have a vague understanding of young and old doesn't mean it can generate CSAM.
Do you actually know that, or are you just assuming it?
Personally, I'm basing my assertions off of experience with related situations, where I've asked image AIs to generate images of things that I'm quite sure weren't in its training set and that require conceptual understanding to create "hybrids." It's done a decent job of those so I'm assuming that it can figure out this specific situation as well, since most of these models have a lot of examples of naked people and young people in their training sets. But I haven't actually asked any AIs to generate images of naked young people to test this one specific case.
My opinion here is that "naked young person" isn't as simple as other compound concepts because there are physiological changes we go through during puberty that an AI can't reverse engineer. Something like "Italian samurai" involves concepts that occur at a surface level that it can easily understand while "naked young person" involves some components that can't be derived simply from applying "young" to "naked person" or "naked" to "young person".
Someone did have a valid counter argument in this subthread though: https://sh.itjust.works/comment/11713795
Well, I haven't gone to any of my image AIs and actually asked them to generate naked pictures of young people. So unless you want to go there this will necessarily involve some degree of theoretical elements.
However, according to the article it's possible to generate this stuff with Stable Diffusion models, and Stable Diffusion models have a negligible amount of CSAM in the training set. So short of actually doing the experiment that would seem to settle it.
I think a lot of people don't appreciate just how surprisingly sophisticated the "world model" that these image AIs have learned is. There was a paper a while back where some researchers were trying to analyze how image generators were working internally, and they discovered that if you were to for example ask one to make a picture of a bicycle it will first come up with a depth map of the image before it starts doing anything to the visual output. That shows that the AI has figured out what the three-dimensional form of a bicycle is based entirely on a pile of two-dimensional training images, with no other clues telling it that the third dimension even exists in the first place.
Just go ask a model to show you, with legal subject matter
Very, very good point. Depending on the answer, I retract the "victimless" narrative.