The Silent Majority: Security Risks of Non-Human Identities in Cloud-Native Applications
Abstract
Cloud-native applications rely on large numbers of credential-bearing software actors, including service accounts, API keys, OAuth clients, pipeline tokens, workload identities and machine certificates. These non-human identities (NHIs) are numerous, frequently over-privileged, often without a named owner, and commonly secured by long-lived secrets that cannot use human-oriented controls such as multi-factor authentication. This paper presents a taxonomy of NHI types and their characteristic risks, and proposes the Discover, Own, Minimize, Replace and Monitor (DOMRM) lifecycle governance model. A four-level maturity model and a set of metrics, such as ownership rate, credential age and ephemeral share, are offered so that organizations can measure progress. A longitudinal evaluation design is outlined. The contribution is conceptual and aims to give practitioners a practical sequence for bringing non-human identities under the same discipline applied to human users.