Models are trained on the same stuff, the responses are not surprising especially when it's not an jailbroken model. dumb article.