DPO Radio

A privacy scanner that surfaces a thousand findings without categorisation is worse than no scanner at all — the privacy team cannot triage the queue and the findings sit unaddressed. The fix is structured categorisation with deterministic severity so the team knows which findings are critical and which are routine.
The privacy-scanner anomaly explanation workflow takes a completed scan and sorts its findings into four buckets: third-party trackers, undocumented endpoints, secure-connection problems, and cookies. Severity is computed from the findings themselves — how many distinct trackers, how many distinct cookie names — rather than being asked of the model, and the scan's overall severity is simply the highest severity of any non-empty bucket. The explanation and the remediation steps are what the assistant writes; the categorisation and the score are not.
The workflow runs against a privacy-scanner finding. The scan source, the offending entity (cookie, tracker, banner config), and the affected operating jurisdiction feed the workflow context.
Third-party trackers, undocumented endpoints, secure connection, and cookies. Each bucket carries its own item list and its own remediation path.
Severity is computed deterministically from data category, consent state, exposure scope, and jurisdiction. The same input always produces the same severity output; the AI does not vary severity assignment by run.
The explanation cites the relevant ePrivacy national overlay and the underlying organisation policy. The DPO sees both the regulatory anchor and the policy reference.
Each finding ships with a suggested remediation routed to the appropriate downstream workflow — banner update, tracker removal, consent flow revision, propagation audit log fix.
The audit trail walks each finding from scanner identification through anomaly explanation and operator remediation routing.



Severity is deterministic; the same input produces the same severity output. The audit trail records the input, the bucket, and the severity for every finding.

The workflow does not remediate autonomously. The DPO accepts the routing; downstream remediation runs through standard workflows with operator confirmation.

Explanations cite the regulatory anchor and the organisation policy reference. Ungrounded explanations ship with the "no grounding" marker.
| Framework | How Anomaly Explanation Supports It |
|---|---|
| EU ePrivacy national overlays | Categorises anomalies against the operating jurisdiction's ePrivacy implementation. |
| EU GDPR | Cites GDPR consent and lawful-basis requirements where the anomaly affects consent state. |

Use scanner findings as input to triage where they relate to an incident. The incident review starts from the relevant recorded facts. See Incident and Continuity Workflows.
Ready to see a queue of 200 privacy-scanner findings categorised in five minutes with deterministic severity? Request a personalised demo.


Run anomaly explanation against your privacy-scanner backlog and measure triage throughput.

Talk to our team about how anomaly explanation fits your scanner cadence.
Third-party trackers, undocumented endpoints, secure connection, and cookies. A scan result is sorted into all four; each bucket reports the number of affected items, its own severity, the affected items themselves, and a remediation path. The scan's headline severity is the highest of any bucket that has items in it.
Yes. Severity is computed from data category, consent state, exposure scope, and jurisdiction. The same input produces the same severity output regardless of run.
No. Findings are categorised and routed; remediation runs through downstream workflows with operator confirmation.
The operating jurisdictions the organisation has configured. The ePrivacy national overlay applied per finding matches the affected jurisdiction.
Yes. Each run sends requests to the AI provider your organisation has configured, and your provider bills your organisation for them directly. A run is never refused because a Forseti allowance ran out, and the Usage page reports what each feature consumed.