It goes beyond me why a corporation with so much to lose does’t have a narrow ai that simply checks if its response is appropriate before providing it.

Wont fix all but if i try this manually chatgpt pretty much always catches its own errors.