> and that appears to be enough to relieve them of any responsibility for the responses.

Unenforceable disclaimers to discourage people from holding you responsible have always existed. "Stay 300 ft back from truck", etc.

With AI, that might be enough of a disclosure, but it might not.