In 2016, London’s Royal Free Hospital (RFH) provided medical records of 1.6 million patients to Google to test a Streams AI app for acute kidney damage. These records, all identifiable patient data, were transferred to Google’s DeepMind and processed as “implied consent for direct care.” The data involved: full names, HIV diagnoses, drug overdoes, abortions, other medical events. The purpose of the Streams AI app was to help doctors intervene and quickly administer treatment, but it is not in use. This month, the National Data Guardian for the UK, Dame Fiona Caldicott, sent a letter to RFH stating that sharing medical data to Google under the guise of implied consent was wrong; the use of patient data to develop new technology is not direct care. If the Information Commissioner’s Office concurs with the assessment, Google must delete the data. DeepMind’s chief claims that the patient data has not been shared with other Google products or used for commercial purposes [kind of scary that this has to be claimed].