The number of Things (Internet of Things devices) is expected to grow to 22 Billion by 2022. These Things are at present notoriously insecure. This presentation will be about using anomaly detection to detect security attacks on Things. This is challenging, especially for applications where thousands or millions of Things are used. I will talk about using statistical time series and machine learning techniques to analyse Things’ normal behavior, and find anomalous behavior, so as to detect when the bad guys are after them.