Privacy Bites AI
Making news around the world this week was the UK’s Information Commissioner Office (ICO) ruling on a contract between Alphabet’s DeepMind and one of the UK National Health Service regional trusts.
Part of a trial to determine the utility of artificial intelligence in the delivery of medical services, the DeepMind system uses a mobile app to help monitor the health of patients with kidney conditions, and then alert primary care providers when there is evidence of a patient’s condition deteriorating. Sounds useful, right? Unfortunately the DeepMind system was given access to 1.6 million patient records without the consent of those patients. Ruled illegal by the ICO, the arrangements have had to be completely redesigned.
An important lesson, no doubt, for anyone using confidential information to feed the insight of new machine learning systems.
Banks across the globe continue to make payouts and multi-billion provisions for mis-selling products to customers over the previous decade. It doesn’t take an advanced engineering degree to work out that misuse of customer records that allow AIs to provide new services or identify new risks could be the source of a new generation of scandals in the corporate or financial world!