In which we look at the White House's Blueprint for an AI Bill of Rights and Google's new method for training LMs with a mixture of objectives.