The warning came from the logs: personal data was moving across trusted identity federation channels.
Identity federation connects systems across organizations using protocols like SAML, OpenID Connect, or WS-Fed. It allows single sign-on and seamless access. But these same pathways can carry Personally Identifiable Information (PII) that should never leave certain boundaries. Detection, in real time, must be precise, fast, and integrated at the protocol level.
PII detection in identity federation is not just about scanning payloads. It means inspecting assertions, tokens, claims, and attributes for names, email addresses, phone numbers, government IDs, or any data defined as sensitive under GDPR, CCPA, or internal policy. This requires parsing formats used in federation—XML for SAML assertions, JSON for OIDC tokens—and applying pattern matching, schema validation, and contextual rules.
The challenge grows when federated systems exchange rich attribute sets. Developers must ensure that mapping rules between identity providers (IdPs) and service providers (SPs) don’t inadvertently leak identifiers or extra attributes. PII detection needs deep integration that can examine decoded tokens and metadata before the handoff completes. Logging alone is not enough.