AI damage detection disclosure
- Version
- 1.1
- Last updated
- 6 August 2026
Notice Draft pending legal review. This document is published for transparency and will be reviewed by counsel before it is relied upon.
This page explains exactly what ValoPal's damage flagging is, so nobody at a stand or in a claim treats it as more than it is.
1.What the feature does
When an attendant photographs a vehicle at drop-off, ValoPal may apply computer vision to those photographs and flag areas of possible existing damage, each with a confidence score. Flags are attached to the ticket alongside the photographs and the attendant's own written note.
2.What it is not
These findings are AI-detected. They are not:
- a vehicle inspection,
- an appraisal or estimate of repair cost,
- a warranty or guarantee of any kind,
- a certification of a vehicle's condition.
No qualified person examines the vehicle as part of this feature. A model looks at photographs and reports what it appears to see in them.
3.The absence of a finding proves nothing
If nothing is flagged, that is not a statement that the vehicle was undamaged. The system reports only what a model observed in the photographs it was given. It can miss damage that is present — because of light, angle, dirt, weather, motion, or the limits of the model — and it can flag damage that is not there.
4.Findings do not determine liability
A finding is a record, not a decision. It does not establish who caused damage, when damage occurred, or who is responsible for it. Responsibility for a vehicle sits with the operator on the terms posted at the stand. Where a finding and an attendant's written note disagree, the attendant's note carries more weight: a person was standing at the car.
5.Photographs and retention
Photographs and their findings are retained for the operator's configured retention period — the ticket close plus the operator's dispute window — and are then deleted along with the rest of the ticket's personal information. See the privacy policy.
6.Training
We do not use customer photographs to train our models. Photographs are sent to our AI vision provider only to produce these findings. We do not grant that provider the right to retain the photographs or use them to train models.
7.How the feature is labelled in the product
Wherever a finding appears — attendant app, manager console, guest pass, exported record — it carries the label "AI-detected — not an inspection". We will not remove that label, and we will not describe this feature as an inspection in sales material.