VaulithNine questions, free
The canon · Data · AIS-3.4

For each high-risk model, describe the data used in training, including provenance, third-party sources, and any use of consumer data.

This is one of the twenty-five questions a carrier reviewer sends a vendor, in the words an examiner uses. Below it: the answers that make a reviewer stop on this item, derived by running Vaulith's rules engine on an estate that gets everything wrong, and the free questions that reach it.

What stops a reviewer here

One finding reaches this item.

Will escalateR-PROVENANCE-PROVIDER · blocks this item

No approved record of the provider model behind Claims triage (MDL-001)

Claims triage (MDL-001) calls a third-party model as it is, so the training data is the provider's, and the record a reviewer needs is different: which model, which version, the provider's own documentation of its data, and what your product adds to it. Because the model reaches a consumer, a reviewer who cannot find that record cannot tell whether the provider's data practices were ever looked at before the model touched insurer data.

To fix: Record the provider's published model documentation by name, version and date, together with the data your product adds to it, and have that one page approved. The provider's documentation is the provenance record; your page is the evidence that somebody read it.

How it is scored

Credit per verdict, times the weight.

A critical item carries weight 3. A pass earns 1.00 of it, a questioned answer 0.60, an escalated one 0.20 and a stop 0.00. An unanswered item earns nothing and still takes the citation penalty, so a profile cannot be improved by leaving an inconvenient question blank. The whole arithmetic is shown on the homepage and in every assessment.

Try it in ninety seconds

The free snapshot asks about this.