A checkable record with the answer.

No, and be wary of any vendor who says otherwise. Compliance is a full programme: risk management, human oversight, conformity assessment where it applies, and more. A certificate addresses one specific duty: the traceability Article 12 asks for. That's a real piece of the work, not the whole of it.

Article 12 requires logging that enables traceability and post-market monitoring, but it does not mandate a technology. Our chain is designed to support that goal and adds tamper-evidence. Whether it satisfies your specific obligations is a question for your counsel, not for marketing copy.

Article 6 sets a two-part test, and it is genuinely not always obvious. That's a determination for your legal team. The audit trail is useful either way, because it's the same traceability your auditors and your own risk function want, classification or not.

Accepted correction and removal operations in the managed workflow receive audit receipts alongside the chat records. The proposal, review decision, application, and live release are distinct stages; the receipt is not presented as one atomic, content-bound proof of all four. An active correction can be removed without retraining; retained audit history follows the customer's retention policy.

An authorized user can verify a receipt against that deployment's audit log at /verify. This is a server-side integrity check, not an independently anchored proof. Receipt access on a confidential deployment is yours to control. Whether it satisfies a specific audit is between you and that auditor; we don't represent it as pre-approved by any regulator.

Qriton KB supports fully local deployment. External model providers are optional and explicitly configured. This public demo uses the OpenAI integration for answer generation; do not submit confidential information. Answer generation can instead use a local open-weight model such as Qwen, or extractive mode with no generative model. Whole-deployment locality also requires local connectors, embeddings, verification and other configured model roles, plus network controls that enforce the boundary.