Security/Safety Improvement Idea

Hi, 

So, you have probably already thought of this but in case you haven't or isn't working on it, maybe this could be a good idea to implement if possible. 

With the recent improvements in AI, we've seen AI models being able to identify complex images, so what if you could implement an AI to "check" every image sent and identify the content to check if it includes bad stuff (CP, promoting self-harm, etc) and can then automatically give a warning/timeout/ban the user if it finds something bad. 

Of course, it might mess it up sometimes, so if the user could appeal a request to get it manually checked, that would be good too. 

I have no idea if any of that is even possible, but anyway, here ya go :)