AI assistants can’t be trusted to produce safe code. The malicious URLs outputted in the response are in clickable form, meaning all a user has to do is click one to be taken to a malicious site.