A sex offender convicted of making more than 1,000 indecent images of children has been banned from using any “AI creating tools” for the next five years in the first known case of its kind.
Anthony Dover, 48, was ordered by a UK court “not to use, visit or access” artificial intelligence generation tools without the prior permission of police as a condition of a sexual harm prevention order imposed in February.
The ban prohibits him from using tools such as text-to-image generators, which can make lifelike pictures based on a written command, and “nudifying” websites used to make explicit “deepfakes”.
Dover, who was given a community order and £200 fine, has also been explicitly ordered not to use Stable Diffusion software, which has reportedly been exploited by paedophiles to create hyper-realistic child sexual abuse material, according to records from a sentencing hearing at Poole magistrates court.
Has your model seen humans in a profile view? Has it seen armor? Has it seen Van Gogh style paintings? If yes then it can create a combo of those things.
For CSAM it needs to know what porn looks like, what a child looks like and what a naked pubescent body looks like to create it. It didn’t make your van Gogh painting from nothing it had an idea of what those things were.
Yes, that’s my point. It didn’t need to be trained on a portrait of Van Gogh in profile; it had several portraits of Van Gogh, a bunch of faces in profile, and used them to create something new. In the exact same way, a network trained on photos of people that include nude adult bodies and children in innocent situations can feasibly create facsimiles of csam without ever having been trained on it.
Yea, specifically, the model shouldn’t have had access to a significant training set on naked prepubescent bodies - that’s been my main objection in this thread.
Except you can’t know that. CSAM has been found in training data already and as long as they pull from social media they will continue to be trained with more.
https://cyber.fsi.stanford.edu/news/investigation-finds-ai-image-generation-models-trained-child-abuse
Awesome link, I’ll share it up thread where someone was asking for it. Yea, it’s something that’s hard to prove since models aren’t upfront with how they’re sourcing their data.
Are you paying attention? It didn’t need to be trained on a portrait of Van Gogh in profile; it had several portraits of Van Gogh, a bunch of faces in profile, and used them to create something new. In the exact same way, a network trained on photos of people that include nude adult bodies and children in innocent situations can feasibly create facsimiles of csam without ever having been trained on it.
The model should not have had access to naked prepubescent imagery. If it did, that’s a problem. My argument in this thread is that it did have access to csam and thus is able to regurgitate them.
I honestly think you and I are in agreement. I’m not arguing that the model is regurgitating known csam but the model ingested csam[1] and the output is derived from that csam. The fact that it can now make csam in the style of Van Gogh is a property of how these models can combine motifs… the fact that it understands how to generate csam at all is the problem.
Ah, I see. I’m sorry; I misunderstood your argument. Yes, given the fact that csam is part of the training data, it would likely be able to reproduce it. I thought your argument was the reverse hypothetical: “If the model is able to produce csam then it must have been trained on csam.” which is incorrect. Again, my apologies for misunderstanding.
The bodies of children are not just small versions of adult bodies.There are meaningful differences that an ai wouldn’t be able to just guess. Also do you not see any problem in using photos of real children to generate csam? Imagine someone used a picture of your child/niece/nephew to generate porn. Does that not feel wrong to you? It’s still using real photos of real children either way, even if it’s abstracted through training data.
I’m discussing hypotheticals of cause-and-effect, not ethics. The question is if it possible not if it’s moral to do so. Please don’t try to shift the topic or try to portray me as possessing an opinion I don’t have again.
While I am aware that there are such differences, I don’t think it’d be impossible for AI to guess them accurately. Lack of training data would make such less probable, since it’d be less likely to know which nude forms better approximate a realistic depiction of the imagined subject. Essentially, certain distributions of outputs have different probabilities depending on if the training data has csam, but due to the diversity of adult bodies it becomes possible for the model to stumble upon a convincing facsimile. How the images of nude adults are labeled can also impact these distributions.
I don’t see a reason to discuss if it’s possible to to something if the thing that’s being done is morally wrong. If you disagree then let’s talk about making a white ethno state or if we can do another Holocaust since morality doesn’t matter when discussing hypotheticals
You can’t generate csam without photos of children to make up the actual child part of the picture. It doesn’t matter if you actually use csam you’re still using photos of children to make pornography. Unless you think ai could create a van gogh style picture without any van gogh training data (and if you do then you don’t know enough about ai generated photos to talk about them with any authority)
What? You can’t think of any reason you might want to discuss whether it’s possible to do something that would be immoral if it is? I don’t believe you’re being honest; I think you’re deliberately trying to deflect because you’ve figured out you’re wrong and don’t want to admit it. Here’s an example to illustrate my point: killing people is generally wrong. Let’s say there’s some discussion about relaxing restrictions on some tool, say knives. Do you really think there’s no point in talking about how it’d be possible to more easily commit murder if such restrictions were relaxed? Discussing the possibility of immoral behavior is an activity that can alter the course of entire civilizations. I cannot fathom how you thought that was a reasonable thing to claim.
Ok. Making a white ethnostate would require committing genocide and forcing all ethnic minorities into a state of subjugation, by definition (unless I’m mistaken). This can theoretically be done without naming these groups in the law. For this reason civil rights are a necessary but not sufficient component in preventing an ethnostate from arising.
There. Happy now? If not, then that’s too bad, because that’s not what I want to talk about.
Sure, but that’s not contradicting my position. Have you stopped disagreeing with me?
If you don’t understand that then I’m done here because you either don’t understand what “ai” does on a fundamental level or you don’t understand how big the difference is between adult and child bodies.
This is a gross conversion to be having on something that is so wrong to do on so many levels.
You can’t make an ethno state without genocide so it is wrong and pointless to talk about
You can’t make ai csam without harming a child so it is wrong and pointless to talk about
Just like how you can’t generate a child without pictures of children to base it on you can’t generate them naked without pictures of their bodies. There is a reason pedos are attracted to those bodies and not women with no curves/small men.
I work with children, I see them everyday. The difference is so massive that an ai would not be able to approximate it with just photos of adults. Ai doesn’t “know” anything it just has photos that it uses to approximate what is being asked based off it’s data. Even if you kept describing in more detail what those bodies looked like it wouldn’t be able to create it without anything to base it on. It’d be like creating a van gogh style picture with no van gogh training data, no matter how much you try to describe the details of his style you’ll never get the ai to make something like it without the training data.
You can keep disagreeing, keep saying “but with more data” but ai can’t make anything original, that is a fundamental misunderstanding of it’s abilities. If it doesn’t have the data it can’t accurately do it.