I might to you. They are not different as they relate to CSAM as a matter of Danish law.
For instance, what if we're NOT talking about CSAM, are generated images ALSO considered equivalent to photos of real things in other domains? Or is CSAM a special carve out and based on what principles? Legal frameworks are stronger when they limit arbitrary rules.
(to be clear, there might be a good logic argument, e.g. it isn't about protecting children, it is about protecting adults - so it doesn't matter how the image came into being).
So, then: What specific and immediate — not potential, but actual — harms does this CSAM-context decision impose on non-CSAM contexts?
At a glance, I don’t see any such harms imposed. One could perhaps argue that AI must take more care now, and that this is a harm; plausible enough, but is that an undue harm? Not likely: the cultural and legal hostility to crossing from non-CSAM contexts into CSAM-context has no reason to yield to the AI operations difficulties imposed by restricting generation of CSAM-context materials.
That’s neither a CSAM-context nor a CSAM-adjacent question as presented, and so is ineligible for the CSAM-context “carve-out” discussions of this post and thread.
However, your question is a boringly incompetent attempt to inflame my emotions in order to trigger illogical responses meant to weaken the impact of my dissection and analysis of this discussion — a particularly ineffective strategy against this high-functioning sociopath. Please don’t use conversational warfare tactics on HN.
The argument that ‘the illegality of AI-generated CSAM would therefore apply to all AI-generated illegality and thus is undue harm upon non-CSAM contexts’ is constructed on the basis that ‘A therefore B’ implies ‘B therefore A’, which is untrue here as detailed above. If you have a better argument in defense of the right to generate CSAM, I recommend a refresher on obscenity laws, how they vary per country, and why they cannot be reduced so simply to ‘CSAM, therefore non-CSAM’ before presenting it; that’s not how it works, no matter how unpalatable or displeasing the one-way discontinuities of CSAM law may be. It’s also worth considering why Apple would both support individual freedoms and privacies while also refusing to deliver the overreach of privacy-destroying CSAM scanning that they at one point had announced last year; that is a very concrete example of undue harm to non-CSAM contexts, and the quality of this discussion would greatly benefit from familiarity in depth with the topic.
The answer to if it generalizes is, as with all new law, that we just don't know. In this case, the procedural for the passing of the last amendment to the relevant section includes the specific point that fictive CSAM should be included. It seems doubtful that other such procedurals would contain the same language. That doesn't rule out that this case is used as a basis for claiming that AI generated images should be treated similarly to real photographs when the two are indistinguishable.
just their "source code" is legaleese.
lots of overlapping concepts even... most contracts have a section where they define all the terms, like defining variables/types. there are clauses and such that guard against specific outcomes, like uhh guard clauses i guess....
even numbered sections like functions to reuse langage.
imean... they even call laws "codes" in places
the one thing that makes it very different is just that they are all stuck with a codebase that has been worked on for 500+ years and its very very hard to actually refactor anything.
also sometimes you dont know if the code will work or what it actually does until its run through the compiler to be executed (courts)
so yea engineers but for a different kind of system.