> I think the issue is that most of the underage content is self published, and not coming from other parts of the web.
That actually doesn't matter; it's straightforward to find the networks of people sharing underage content. It is very rare that someone will only be sharing fresh, self-published content which is never shared or reposted by anyone else who also shares content in the ICMEC database at any point.
If you're analyzing content in a sheer vacuum, yes, this would be a problem, but you're not - you're analyzing content which is spread across a social network, and you're looking for networks which are intentionally seeking out and sharing this material. It is a much more straightforward problem than most people realize; it just requires the right staffing and resources (ie, problems which are directly solved by money).
> This makes checking against the ICMEC database worthless, as the content is brand new and has never been online before.
On the contrary; that's the exact mechanism by which the ICMEC database gets expanded.
This is why even the DOJ opposed SESTA and FOSTA - websites that are used by sex workers are also great places to find and rescue victims of sex trafficking. Most sex workers aren't trafficking victims, but a large number of trafficking victims are listed on those sites as well. It was one of the primary ways that the DOJ would rescue victims and prosecute traffickers. With those sites gone, traffickers have gone underground, and it's much more difficult to find them.