Utilizing canary data to identify improper data access

Abstract

This disclosure describes techniques implemented partly by a service provider to monitor a cloud-based service by generating and placing canary records in storage locations along with real records to identify improper access events of the records or other data. The service provider may detect an access event where records in a storage location were accessed, and determine whether a canary record was accessed. If a canary record was accessed, the service provider may determine that the access event was potentially performed by a malicious entity because authorized users generally may not have reason to access a canary record when utilizing their cloud-based service. The service provider may generate canary records that are difficult to identify by a malicious entity, and may position canary records in the storage locations to help ensure that the canary records are accessed by a malicious entity during an improper access event.