No automated decision about a citizen without a human and an appeal

The problem

Automated systems are increasingly used in decisions about entitlements, and affected citizens often cannot find out why a decision was made.

Why it matters

An unexplainable denial of an entitlement is indistinguishable from an arbitrary one.

Proposed solution

Advocate that any automated system materially affecting a citizen must have a named human decision-maker, a stated reason, and a working appeal route.

How it would be implemented

We hold ourselves to this first: every AI output in this platform is advisory, recorded with its model version, and reviewable.

Risks, and what we would do about them