[ad_1] AI companies claim to have robust safety checks that ensure models don’t say or do strange, illegal, or unsafe ...