Enkrypt AICopy wiki link

Enkrypt AI sells guardrails and governance controls for model and agent workflows. Its documented detectors cover areas such as prompt injection, toxicity, and personally identifiable information (PII), while its agent controls can approve, modify, or block tool behavior.

Guardrails reduce selected risks; they do not create a cryptographic confidentiality boundary. A hosted inspection or gateway service receives the content it evaluates unless a separately verified deployment model changes that path. False positives, false negatives, and bypasses also require application-level controls.

Private AI competitors therefore treats Enkrypt AI as a policy and security layer rather than as private inference itself.