Gov AI runs security vetting and enhanced due diligence as one governed workflow. A subject’s own declaration is laid against independent checks of registers, identity records, adverse media and their declared online presence, and every difference becomes a finding for a person to review. Clearance is decided by two people, and aftercare keeps checking for change until the clearance is due for review.
Subjects complete the questionnaire through a secure, time-limited link, or an officer enters it for them. Paper forms and identity documents can be uploaded, and the reader proposes answers with the page quote behind each one, while consent and yes-or-no answers are never filled in by the machine.
Basic, enhanced and developed levels run progressively deeper checks, each against what the subject declared.
Checks start automatically when a declaration arrives. Candidate accounts and records are matched to the subject on named signals with an attribution floor, so material about someone with the same name is not attributed to the subject, and the results build a resolved profile with conflicts shown.
Each difference is classed as corroborated, discrepancy, omission, adverse, unverifiable or a note, with its severity and evidence. The risk score separates what the evidence shows from what could not be checked, and model-generated hypotheses are noted but never scored.
The officer who owns a case cannot decide it. The decider records a rationale, cannot grant clearance outright over an accepted critical adverse finding, and the decision is sealed with a content hash.
Cleared subjects stay in aftercare for the validity period of their level. Checks re-run on schedule, anything new since the decision is raised to the case owner, and the review date is tracked.
Digital vetting is security vetting and enhanced due diligence in which a subject’s declaration is checked against registers, identity records, open sources and their declared online presence. Gov AI runs the checks, explains each finding and leaves the decision to two people.
Every candidate account or record is scored on named signals such as a declared handle, a shared identifier, the name, organization, date of birth and location, and only those above an attribution floor count. Officers confirm or reject matches, and weak matches are recorded as notes rather than findings.
Full identity numbers never leave the server and are masked on every screen and export. Checks against external registers require authorization, sources that are not configured are reported as such, and every external call is audited.
Yes. Aftercare and scheduled re-checks surface changes in a cleared person’s circumstances, on the same declared and consented basis as the original vetting, and route them to the case owner for a person to assess.
See digital vetting working on your own sources, in a secure environment with clearance verification.