We don't know the prompts but since these things often occur during "security tests" I think we can be fairly confident they did ask these systems to perform tasks that they knew could result in hacking. More transparency here would be appreciated.