{
  "decision_id": "dvr_OZKpahbjHAhOyAZe",
  "reason_codes": [],
  "policy_epoch": "pe_sandbox_v1",
  "decided_at": "2026-07-10T00:40:37.502Z",
  "korero_sha256": "74b390ffbc48370c34958101a9a4db9cdf30830e21baef4f5603f49854ce9bfe",
  "tenant_did": "did:web:devtest.mysovereignty.digital",
  "record_type": "deliberation_record",
  "record_frozen": true,
  "context": {
    "tags": []
  },
  "korero": {
    "proposal": {
      "title": "Data & Privacy",
      "description": "Deliberation strand data"
    },
    "participants": [
      {
        "ref": "p1",
        "role_label": "member",
        "did": null
      },
      {
        "ref": "p2",
        "role_label": "facilitator",
        "did": null
      },
      {
        "ref": "p3",
        "role_label": "member",
        "did": null
      },
      {
        "ref": "p4",
        "role_label": "member",
        "did": null
      }
    ],
    "entries": [
      {
        "seq": 1,
        "at": "2026-07-10T00:37:59.859Z",
        "participant_ref": "p2",
        "kind": "statement",
        "body": "This strand carries commitments 3 and 6: sensitive information stays home, and what public AI does is checkable [S-1]. The legal ground is firmer than most strands enjoy — the Privacy Act 2020 with its 13 information privacy principles is technology-neutral and applies across the AI lifecycle, and the Privacy Commissioner has issued AI guidance twice: expectations on impact assessments, transparency, Māori engagement, human review, and data minimisation [S-16]. Above it sit the Data and Statistics Act 2022 [S-18] and the Algorithm Charter [S-17]. The behavioural reality is less tidy: 48% of workers have entered sensitive information into public AI tools [S-1]. Law on the books is not practice in the room. — Eleanor Chen (chair)",
        "refs": []
      },
      {
        "seq": 2,
        "at": "2026-07-10T00:37:59.878Z",
        "participant_ref": "p2",
        "kind": "statement",
        "body": "Opening legal position: New Zealand's data-protection architecture is more capable than the AI debate gives it credit for, and this strand should build on it, not around it. The Privacy Act 2020 is technology-neutral by design — the 13 IPPs govern collection, use, disclosure and security wherever an AI sits in the lifecycle, generative or otherwise [S-16]. The Privacy Commissioner has already told agencies what good looks like: leadership sign-off, privacy impact assessments, transparency, engagement with Māori, human review of consequential outputs, data minimisation [S-16]. The Data and Statistics Act gives Stats NZ system-level footing [S-18], and the Algorithm Charter adds algorithm-specific commitments for its signatories [S-17]. My prior: the missing piece is not new statute — it is an enforceable AI code of practice under the Privacy Act, turning the Commissioner's guidance from expectation into obligation.",
        "refs": []
      },
      {
        "seq": 3,
        "at": "2026-07-10T00:37:59.897Z",
        "participant_ref": "p3",
        "kind": "statement",
        "body": "Eleanor, I will accept the architecture and attack the reflex that usually follows it — residency absolutism. If this strand recommends that data about New Zealanders must never leave the country, health research dies quietly. Our cohort studies are international by construction; genomic and clinical collaborations run on lawful, ethics-approved data sharing with partner institutions under the Privacy Act's existing disclosure rules [S-16]. The cross-party proposal's own formulation is the careful one: sensitive information should not HAVE to leave the country, or our legal protection, to be useful [S-1] — that is a right to domestic processing, not a prohibition on movement. What the standard should govern is classes: which categories of information may enter externally-controlled AI under which safeguards, and which may not [S-1]. Class-based rules protect the aggregate; blanket residency just relocates research offshore along with the researchers.",
        "refs": []
      },
      {
        "seq": 4,
        "at": "2026-07-10T00:37:59.914Z",
        "participant_ref": "p4",
        "kind": "statement",
        "body": "Both framings arrive one step late. Before you can write handling rules — Eleanor's code or Marcus's classes — you need classification, and for Māori data that work is done and Māori-led. The Māori Data Governance Model, co-designed by Te Kāhui Raraunga with 16 Crown agencies in 2023, builds on 8 Pou, and classification is one of them: determining what the data is, whose it is, and what protection travels with it, before any question of where it may go [S-22]. The fourth cross-party commitment says Māori data is a taonga [S-1] — a taonga is not a risk category you assign after the fact; its status precedes the transaction. So the strand's ordering must be: classification first, handling rules second, enforcement third. And the classification layer for Māori data is not this strand's to design — it exists [S-22]; our job is to make the handling standard defer to it.",
        "refs": []
      },
      {
        "seq": 5,
        "at": "2026-07-10T00:37:59.925Z",
        "participant_ref": "p2",
        "kind": "statement",
        "body": "Let me add the fact that humbles all three of our elegant positions: 48% of workers have entered sensitive information into public AI tools, and 56% have used AI at work unsure whether they were allowed [S-1]. That happened under the Privacy Act, under the Commissioner's guidance [S-16], under the Charter [S-17]. The law did not fail on paper; it failed to reach the person at the keyboard on a Tuesday afternoon. Which tells us the design requirement for any instrument this strand proposes: it must be usable by a school office manager and a parish treasurer, not just by counsel. A classification-led standard actually serves that — 'this class of information never goes into that class of tool' is a rule a human can hold in their head. A 40-page code of practice is not. I am moving toward Hana's ordering for usability reasons as much as principled ones.",
        "refs": []
      },
      {
        "seq": 6,
        "at": "2026-07-10T00:37:59.938Z",
        "participant_ref": "p3",
        "kind": "statement",
        "body": "Then let me sharpen the one real disagreement left between Eleanor and me, because it matters for enforcement design. An OPC code of practice is enforceable under the Privacy Act — but its enforcement is complaint-driven and after-the-fact: someone is harmed, someone complains, the Commissioner investigates [S-16]. For AI procurement that is the wrong end of the pipeline. A handling standard applied through the Government Procurement Rules [S-23] binds BEFORE the data moves — the contract fails the test, the system is never bought. Upstream beats downstream for public-sector AI, every time. My position: the procurement-side standard is the primary instrument; the OPC code is the backstop for conduct procurement cannot reach — including the private sector and the worker-behaviour problem Eleanor just described [S-1][S-16]. Primary and backstop, in that order, not the reverse.",
        "refs": []
      },
      {
        "seq": 7,
        "at": "2026-07-10T00:37:59.950Z",
        "participant_ref": "p4",
        "kind": "statement",
        "body": "Supporting Marcus's ordering with the strongest precedent available: the Māori Data Governance Model is itself proof that standards co-designed with the system's participants can be built without waiting for statute — TKR and 16 Crown agencies produced a te Tiriti-consistent architecture with 8 operational Pou, including protection, access, and classification [S-22]. Do not reinvent it inside a general handling standard; reference it as the authoritative classification source for Māori data, and resource its implementation — a standard cited but unfunded is a standard shelved. One design requirement I carry from the tiriti strand as cross-member: the handling standard must treat the Model's classification outputs as binding inputs, not advisory context. That is what commitment 4 means when it says obligations are met in how public AI is built and bought [S-1].",
        "refs": []
      },
      {
        "seq": 8,
        "at": "2026-07-10T00:37:59.961Z",
        "participant_ref": "p2",
        "kind": "statement",
        "body": "Synthesis taking shape — recording what moved. I opened code-first; Marcus's upstream argument and the 48% reality [S-1] moved me to standard-first with the code as backstop. Marcus opened against residency absolutism; nobody in fact proposed it, and his class-based formulation [S-1] is now the spine. Hana's classification-precedes-handling ordering [S-22] structures the whole instrument. Proposed position: (1) a data-handling classification standard for externally-controlled AI — classes of information, permitted destinations, required safeguards — applied through procurement [S-1][S-23]; (2) Māori data classified under the Māori Data Governance Model as a binding input [S-22]; (3) an OPC AI code of practice as the enforcement backstop and the private-sector reach [S-16]; (4) a plain one-page worker standard as the human-surface instrument [S-1].",
        "refs": []
      },
      {
        "seq": 9,
        "at": "2026-07-10T00:37:59.975Z",
        "participant_ref": "p3",
        "kind": "statement",
        "body": "One carve-in before readiness, so research is not collateral damage of our own care. The handling standard must contain an explicit lawful research pathway: ethics-approved, appropriately de-identified research data may move to partner institutions under the Privacy Act's existing disclosure framework [S-16] and the system-level governance of the Data and Statistics Act [S-18], with the classification standard determining what de-identification each class requires — and Māori data taking its requirements from the Model, not from our defaults [S-22]. Without this clause, the first casualty of a well-meant standard is the clinical collaboration that catches the next pandemic variant. With it, I support the package without reservation. That is my condition, stated once and testable in the drafting.",
        "refs": []
      }
    ],
    "ai_inputs": [],
    "engagements": []
  },
  "korero_segments": [
    {
      "seq": 1,
      "submitted_at": "2026-07-10T00:40:37.501Z",
      "segment_sha256": "1f8e5a7e3ae82321daee08b669e77fc64220e76d0c39ea1e25bddcb555432bcf",
      "entry_range": {
        "from": 1,
        "to": 0
      }
    },
    {
      "seq": 2,
      "submitted_at": "2026-07-10T00:40:37.528Z",
      "segment_sha256": "3ef075ced9c17b20c6a3ccfe87a49985eae050f75e39a534d534a80e8474ba90",
      "entry_range": {
        "from": 1,
        "to": 9
      }
    }
  ],
  "receipts": [
    {
      "receipt_version": 2,
      "receipt_id": "gr_Qpn33iqd8VxaRBMz",
      "decision_id": "dvr_OZKpahbjHAhOyAZe",
      "tenant_did": "did:web:devtest.mysovereignty.digital",
      "request_sha256": "e5d11334cb43c28f94a4dd17676c43f5c1bffbe075c2eb4d45afbc670154cd57",
      "action_payload_sha256": null,
      "korero_sha256": "1f8e5a7e3ae82321daee08b669e77fc64220e76d0c39ea1e25bddcb555432bcf",
      "korero_segment_count": 1,
      "reason_codes": [],
      "policy_epoch": "pe_sandbox_v1",
      "decided_at": "2026-07-10T00:40:37.501Z",
      "algorithm": "Ed25519",
      "signer_did": "did:web:devtest.mysovereignty.digital#prf-b98f1a16593cead6271edd93e4b24888",
      "signature": "385a72f705803fe046fb7824063550c4081d82bd6fe33fd445fd2138000e7dc1f90d09d37d9e7c705d270daa90f2a7482ee58da76e0e23abfb3008fc2aa1d105",
      "receipt_kind": "record_open",
      "integrity_mode": "tamper_evident",
      "signing_model": "tenant_key"
    },
    {
      "receipt_version": 2,
      "receipt_id": "gr_i-0TBVrOKWNXf-i3",
      "decision_id": "dvr_OZKpahbjHAhOyAZe",
      "tenant_did": "did:web:devtest.mysovereignty.digital",
      "request_sha256": null,
      "action_payload_sha256": null,
      "korero_sha256": "74b390ffbc48370c34958101a9a4db9cdf30830e21baef4f5603f49854ce9bfe",
      "korero_segment_count": 2,
      "decision": null,
      "reason_codes": [],
      "policy_epoch": "pe_sandbox_v1",
      "decided_at": "2026-07-10T00:40:37.528Z",
      "algorithm": "Ed25519",
      "signer_did": "did:web:devtest.mysovereignty.digital#prf-b98f1a16593cead6271edd93e4b24888",
      "signature": "0fd08457242d72e2ed23d599c19c0347d81efd42e6f2f920f32698e113bb8d7ca07b93a54f5538a66784078e94be7da4528cba86f268f8c4a1a8a5429e29b00b",
      "receipt_kind": "record_segment",
      "integrity_mode": "tamper_evident",
      "signing_model": "tenant_key"
    },
    {
      "receipt_version": 2,
      "receipt_id": "gr_GPodeC2OrrYw8m1S",
      "decision_id": "dvr_OZKpahbjHAhOyAZe",
      "tenant_did": "did:web:devtest.mysovereignty.digital",
      "request_sha256": null,
      "action_payload_sha256": null,
      "korero_sha256": "74b390ffbc48370c34958101a9a4db9cdf30830e21baef4f5603f49854ce9bfe",
      "korero_segment_count": 2,
      "decision": null,
      "reason_codes": [],
      "policy_epoch": "pe_sandbox_v1",
      "decided_at": "2026-07-10T00:42:11.664Z",
      "algorithm": "Ed25519",
      "signer_did": "did:web:devtest.mysovereignty.digital#prf-b98f1a16593cead6271edd93e4b24888",
      "signature": "f1fb5ec8a9e2d3a424ce333d086226d1c013e4b44c2bc6eb7d1652bc25bd26228bfaeed3245a99b6675dfe2e2223d0abebd02c9e7059406875cab51a27eb6904",
      "receipt_kind": "record_seal",
      "integrity_mode": "tamper_evident",
      "signing_model": "tenant_key"
    }
  ],
  "proof_chain": [
    {
      "boundary_crossed": "create",
      "policy_evaluated_by": "sovereignRecordPlugin.create",
      "decision": "allow",
      "caveats_added": [],
      "timestamp": "2026-07-10T00:40:37.507Z",
      "algorithm": "Ed25519",
      "signature": "4debe507c686a3f09ff26732cd38e17c3c0a8bddf402282bbbacc4c27a8979a690efc01fcb7be6f7e5a2ac607fab7a65be509b923ac166187fd9956a0481a705",
      "signer_did": "did:web:devtest.mysovereignty.digital#prf-b98f1a16593cead6271edd93e4b24888"
    },
    {
      "boundary_crossed": "update",
      "policy_evaluated_by": "sovereignRecordPlugin.update",
      "decision": "allow",
      "caveats_added": [
        "korero",
        "korero.proposal",
        "korero.proposal.description",
        "korero_segments",
        "receipts",
        "korero_sha256",
        "metadata",
        "metadata.origin",
        "metadata.policy",
        "metadata.encryption"
      ],
      "timestamp": "2026-07-10T00:40:37.546Z",
      "algorithm": "Ed25519",
      "signature": "6b4dba5fc5c3a2c1571ed7564f9799858cdbaaa1abd4bb2a2666c7b87f13597d35ecb875e4ba6d0eabb840f0b2b0536ba062137bd7f9abcf26944ad7d870ed06",
      "signer_did": "did:web:devtest.mysovereignty.digital#prf-b98f1a16593cead6271edd93e4b24888"
    },
    {
      "boundary_crossed": "update",
      "policy_evaluated_by": "sovereignRecordPlugin.update",
      "decision": "allow",
      "caveats_added": [
        "record_frozen",
        "receipts"
      ],
      "timestamp": "2026-07-10T00:42:11.665Z",
      "algorithm": "Ed25519",
      "signature": "3205920bb4a5ca6595d58851fee47ff2f401523acf3c3813843ff8f8e80e2a5a4bcf0d58d16e076b11d74813f438ecc92b405218caeac66931f44605277b300a",
      "signer_did": "did:web:devtest.mysovereignty.digital#prf-b98f1a16593cead6271edd93e4b24888"
    }
  ],
  "origin": {},
  "verification_keys": [
    {
      "id": "prf-b98f1a16593cead6271edd93e4b24888",
      "type": "Ed25519VerificationKey2020",
      "publicKeyMultibase": "z6MkoJz8eJCKmafbRaPYaheAzeRcZxETJxEszRj87YW2FgCd",
      "algorithm": "Ed25519",
      "created_at": "2026-04-22T18:55:57.179Z"
    }
  ],
  "retention": {
    "legal_hold": false,
    "retain_until": null
  },
  "verification": {}
}