I'm so sorry you went through this.

I'm sorry you had to handle that. Though, this is an inevitable problem for a site hosting user-generated content of any sort. I think not hosting your own content is the biggest first step. It might be better to depend on external image/filesharing sites like how things used to work with imgur, photobucket, flickr, etc. (is photobucket around anymore, lol?). This does pose a risk for link rot and what not, but I think given the scale of this operation, there's no reason to having to be scanning your own servers for illegal content any more than you have to for basic moderation. I'm not sure how this issue works with federation (if another lemmy instance hosted a bad image/thumbnail/message, does that get copied over to beehaw?), but if your risk tolerance doesn't want to deal with it, a non-federated option would give you more control.

In general user generated content is always going to pose a risk to the website hosting it. It's a matter of good risk management, from prevention and mitigation to an effective response, that will best serve both the administration and the users in terms of ensuring a safe service and minimizing legal risk.

Aren't there ways to automatically scan the content of posts to detect csam and other disturbed content?

Because completely disallowing images isn't the solution. For once, people can still share those things through links. And secondly, there are people who need images and the like to communicate.