SCIM 2.0 deprovisions a person by setting active to false on a User resource, and every downstream system watches that flag. An AI agent breaks the model in a specific way: the agent is usually not a User in the…
First seen on securityboulevard.com
Jump to article: securityboulevard.com/2026/09/scim-agent-provisioning-9-failures-when-the-principal-is-not-a-person/
