GDPR¶
Statement¶
ClaimGuard treats GDPR as the baseline standard for personal-data handling across the product, even where customer contracts do not specifically require it. The technical foundations — encryption at rest and in transit (where it exists), role-and-org access control, audit logging, defined subprocessors, EU-region storage, a clear incident-response intake — are in place and described elsewhere in this portal. The paperwork layer (DPA template, Records of Processing Activities, formal DPIA process, lawful-basis matrix per processing activity) is not yet formalized. That is the gap that keeps this control at partial.
Implementation¶
What we are, GDPR-shape¶
- Role. ClaimGuard is a processor for customer-controlled personal data (claim data, evidence, end-user PII the customer uploads). Customers are the controllers.
- Subjects. Customer staff (users of the application) and any individuals whose data appears in claim narratives or evidence files.
- Establishment. ClaimGuard's primary cloud footprint is in
europe-west1(Belgium) — data at rest lives in the EU.
What's in place technically¶
- EU storage for all primary application data and snapshots. See Cloud provider and Backups.
- Encryption at rest under GMEK across disks, snapshots, GCS, and Secret Manager. See Encryption at rest.
- Outbound TLS for cloud-control-plane and AI-vendor traffic. See Cryptography.
- Role + org access control with org-scoping at the SQL layer, giving a customer-controller-style isolation between tenants. See Authorization.
- Cloud audit log with 400-day immutable retention covering privileged access. See Audit logging (cloud).
- Documented subprocessors with EU-residency notes. See Subprocessors.
- Vulnerability disclosure path for external researchers and a documented incident-response lifecycle (partial). See Vulnerability disclosure and Incident response.
What's not in place yet¶
- No published DPA template. A DPA can be negotiated bilaterally on request today, but there is no standard template hosted under this portal.
- No Records of Processing Activities (ROPA) document distinct from the data-classification page. ROPA is the formal artifact GDPR Art. 30 expects from a processor of any non-trivial scale.
- No formal DPIA process triggered by, e.g., an AI integration introducing automated decision-making elements (the AI today is advisory, not automated decision-making, so this is not currently triggered — but the process to evaluate it should exist).
- No published privacy notice customer-facing on the trust portal. Cross-listed on Privacy notice (planned).
- No 72-hour breach-notification runbook codified per Art. 33. Cross-listed on Incident response as a known gap.
- No data-subject-request (DSR) workflow for requests routed through ClaimGuard rather than through the customer-controller. Today such requests would be passed to the controlling customer.
- No formal lawful-basis matrix documenting, per data class, the legal basis our customers can rely on (their contractual necessity, our legitimate interest as processor, etc.).
Data-subject rights¶
For end users whose data is in a customer's tenant, ClaimGuard fulfills rights via the customer-controller:
- Right of access / rectification / erasure / restriction: the customer-controller invokes the relevant application endpoints (CRUD on claims and evidence) on the data subject's behalf. ClaimGuard supports the controller in fulfilling the right; we are not the first contact for the data subject.
- Right to data portability: the customer-controller can export claim and evidence data via the application UI / API.
- Objection / automated decision-making: the AI integration is advisory; a human reviewer makes determinations. See AI transparency.
For ClaimGuard's own users (customer staff who hold the application account), data-subject requests can be routed to security@dtectvision.ai and are handled per the same SLA as a vulnerability report.
International transfers¶
Data at rest is stored in europe-west1. Two outbound flows leave
the EU boundary:
- Google Gemini API — managed by Google globally; covered under the Google Cloud DPA. See Subprocessors.
- Cloud-control-plane traffic to GCP services — Google's standard EU-EEA / SCC posture applies as defined in the Google Cloud DPA.
A stronger residency commitment (e.g., "no inference outside the EU") would require a Gemini API plan change and is a roadmap item under AI transparency.
Status¶
partial — verified 2026-04-29.
What's in place:
- EU-region data residency for primary storage and backups.
- Encryption at rest and the rest of the technical control set.
- A documented subprocessor list.
- An incident-response intake; a vulnerability disclosure path.
- Honest data-flow disclosure for the one outbound AI integration.
Known gaps¶
- No published DPA template on this portal.
- No formal ROPA distinct from the data-classification page.
- No DPIA process documented.
- No 72-hour breach-notification runbook codified.
- No DSR workflow for ClaimGuard-routed requests.
- No lawful-basis matrix per processing activity.
- No published privacy notice.
Roadmap¶
- DPA template as a downloadable PDF on this portal.
- ROPA as a maintained internal doc with a quarterly review.
- DPIA process (a one-page checklist + decision record) before any new AI surface or any new data class lands.
- Breach-notification runbook with a documented 72-hour timeline and named comms owner. Cross-listed on Incident response.
- DSR workflow routed through
security@dtectvision.aiwith an initial-response SLA matching the VDP. - Lawful-basis matrix per data class.
- Public privacy notice at Privacy notice once drafted.