> Nowadays if you share or reproduce anything controversial, is as if you are endorsing such content. LLM's outputs are user generated content for what is worth

> Nowadays if you share or reproduce anything controversial, is as if you are endorsing such content. LLM's outputs are user generated content for what is worth it, both from inferece and from training. That line of thinking that the host would be responsible for everything it says, even if the users force it to, is ludicrous to me but it seems like how the world nowadays has become. If it applies to Facebook, Twitter, YouTube, ... etc. it will apply to AI chatbots. Facebook didn't write those posts that radicalize, promote hatred, and ignite conflict, but if it publishes them and promote them to users, the FBI will want them to take it down.