FOUNDING MIRROR CITY — Applications are open for founding participants, research sponsors and public-interest partners. Explore the Founding Pilot →

PRIVACY, AUTHORITY AND EVIDENCE

TRUST REQUIRES ARCHITECTURE.

CivicLedger.ai is designed around one principle:

The person should not have to surrender control of their political identity in order to participate in civic intelligence.

Three things that must remain separate

Eligibility is not identity. Identity is not opinion. Opinion is not permission.

Eligibility layer

Confirms only the attributes necessary for a particular Civic Mission, such as:

  • unique eligible participant;
  • age threshold;
  • jurisdiction;
  • residency assurance;
  • credential validity.

Where possible, CivicLedger.ai should receive a verified claim rather than a copy of the underlying identity document.

Private Civic Agent layer

Contains the user-controlled profile:

  • values;
  • priorities;
  • interests;
  • standing instructions;
  • private response history;
  • personal civic reports.

Mission response layer

Contains only the information required for a specific Civic Mission, together with the permissions and retention terms accepted for that mission.

The institutional buyer receives only the approved aggregate output.

Blockchain boundary

Proofs on-chain. Political profiles off-chain.

Where cryptographic anchoring adds value, CivicLedger.ai may record:

  • hashes of public source versions;
  • consent-disclosure versions;
  • private participant receipts;
  • one-participation proofs;
  • tally commitments;
  • public-report versions;
  • payout-pool records.

CivicLedger.ai should not publish on a public chain:

  • political profiles;
  • individual responses;
  • precise residential information;
  • identity documents;
  • stable identifiers connecting a person across every issue;
  • private Civic Agent instructions.

The ledger is the evidence layer.

It is not an eternal political dossier.

Buyer restrictions

Institutional access does not include:

  • raw individual rows;
  • exact participant locations;
  • direct contact information;
  • re-identification rights;
  • cross-mission political profiles;
  • campaign targeting;
  • resale of participant information;
  • training unrelated third-party models on personal response data without new permission.

AI boundaries

The AI must show its work—and its uncertainty.

Civic Agent outputs should distinguish:

  • official fact;
  • quoted or summarized source content;
  • analytical inference;
  • forecast or potential consequence;
  • missing evidence;
  • confidence level;
  • user-supplied value judgment.

The agent does not become correct merely because its answer is fluent.

Trust closing

No invisible sponsor. No invisible permission. No invisible authority.

[Contact the Privacy Lead]