Taonga in the Latent Space
Yann LeCun left Meta to bet that AI’s missing piece is a model of the world. The first world worth modelling is a community: the Village already holds the signed record of a community’s state and the gate that checks an action before it runs — what’s missing is the picture that lets an AI foresee what its action would do to the people. A community-state model, with the community as kaitiaki of the model of its own world, taonga and consent first-class, federated not concentrated. Proposed as the next step, not yet shipped.
Earning the Right to Propagate
How a human Village governs its own AI. Vincent Boucher (QUEBEC.AI & MONTREAL.AI) published a constitution for governing how institutional AI may evolve — GoalOS: The Proof-of-Evolution Constitution — and most of its primitives were ones the Village had already shipped for human governance. We read the convergence as external validation, adopt the proof-gated promotion discipline we were missing, refuse the blockchain the standard runs on (on sovereignty grounds its own neutrality clause permits), and add the gate it lacks: keep the community the author. AI is the servant, not the subject.
Governance That Can’t Be Quietly Undone
NZ and Australia chose principles over prescriptive AI law. The Village makes those soft-law principles — transparency, human oversight, auditability, data sovereignty — structurally enforced, and runs real governance-village and kāhui Māori board deliberations, meetings, and votes on a tamper-evident substrate. Shipped vs in-development kept distinct.
Sovereignty Without Dominance
Rightful authority at human scale, in the age of AI. Extends Graylin rather than rebutting him, asks Western readers to scrutinise the US as evenly as they scrutinise China — with respect for every party — and holds that sovereignty is not the capacity to dominate a technology but a community’s rightful authority over the systems that act on it: federated, never surrendered. Closes with a live exhibit from the days it was written.
Kaitiaki Intelligence and Mokopuna Recorder
Two prototype briefs and a reflection, from a consciously limited non-Māori standpoint. Western AI debate leads with “is it conscious?”; te ao Māori asks instead what relations an entity inhabits, whose authority it answers to, and whether it strengthens the mauri around it. Two small forms carry the argument — a place-based kaitiaki intelligence for one taonga, and a mokopuna recorder for whānau memory — where the AI assists and the people decide.
Federate, Don’t Align
The operational companion to The Map Has No Node for Legitimacy. If a small nation will never win the AI capacity race, what is the lowest-risk way to hold the authority it can? Read as a risk decision, three options appear — align with the American stack, align with the Chinese stack, or federate and align with neither. Only the federated mesh — of communities and of their inference — carries no irreversible tail. Extends to a national-scale federated Aotearoa, and names the posture: a non-aligned layer for AI.
The Map Has No Node for Legitimacy
A response to Tim Clancy and Asmeret Bier Naugle’s qualitative model of AI sovereignty. Their five-lever framing measures sovereignty as capacity — and has no node for legitimacy, for the rightful authority of the people whose data trains and steers a system. Governance sovereignty and substrate sovereignty are separable; rightful authority is the layer the Village platform and Tractatus framework run at today.
Before it acts, can the AI picture what it will do to you?
The plain-language companion to Taonga in the Latent Space. LeCun says today’s AI can’t picture the consequences of its actions. The first place to give it that picture is a community — and the community must be the guardian of the picture. The world model is how it pictures; the community is who decides.
When does an AI earn the right to replace the one you already trust?
The plain-language companion to Earning the Right to Propagate. A new constitution for governing how AI may improve turns out to describe most of what we’d already built for human communities. We agree with almost all of it, refuse the blockchain it runs on, and add the one gate it’s missing: keep the people the authors. AI is the servant, not the subject.
Soft Rules, Hard Governance
The plain-language companion to Governance That Can’t Be Quietly Undone. NZ and Australia didn’t make AI laws — just guidelines; here’s how a community can build the good behaviour into the architecture so it can’t be quietly undone.
Someone Else’s Switch
The plain-language companion to Sovereignty Without Dominance. Last week the US government switched off the writer’s best AI tool overnight — and that is the whole argument: if you build your community’s life on someone else’s stack, they hold the switch. A general-readership case for owning the AI that runs your community’s life.
AI That Knows Its Place
The plain-language companion to Kaitiaki Intelligence and Mokopuna Recorder. Stop asking whether AI is conscious; ask what it’s made from, who it answers to, and who gets to say no — with a seven-slide presentation view.
Federate, Don’t Align
Federated communities and federated inference as the lowest-risk AI option for small nations — the operational, risk-priced companion to The Map Has No Node for Legitimacy, with a ten-slide presentation view.
Sovereignty Isn’t Only Something You Can Buy
The legitimacy argument in short form: AI sovereignty as rightful authority, not only the capacity you can buy. A response to Clancy and Naugle, and why the Village runs at the authority layer.
Village and agentic AI: where we are, and what the next step would look like
A reading of the architectural and framework papers against the actual Village implementation. Two diagrams on identical frames map the current platform and the next step toward a genuinely agentic runtime surface, so the difference reads as a small addition rather than a redesign. The argument the diagrams together demonstrate: the dominant agentic stack's optimisation choices produce structural outcomes that institutions whose civic mediation should remain self-governable have grounds to refuse, and an inverse substrate is buildable.
A Civil-Society Proposal for Sovereign and Federated Agentic AI in Aotearoa New Zealand
A civil-society contribution to NZ policymakers and community organisers, structurally mirroring the People’s Republic of China’s 2026 Implementation Guidelines on Intelligent Agents. Six sections, 14 sub-sections, 38 numbered items, with a new §0 “Philosophical Foundations” chapter prepended drawing on the Tractatus framework, the CARE Principles for Indigenous Data Governance, Te Mana Raraunga and Dr Taiuru’s published scholarship, and ISO/IEC JTC 1/SC 42 standards. Proposes a single committee under a suitable umbrella organisation with five named workstreams whose principal product is contribution to ISO/IEC SC42 international standards work and bilateral dialogue with the CAC framework’s authors.
Sovereign-Record Architecture for Community-Scale Platforms
A research paper reporting an alternative substrate for community-scale platforms, in which the sovereignty a community needs is a property of the records themselves rather than a concession the operator may revoke. Cryptographic provenance, tenant-bounded policy enforcement, bilateral and bounded federation, member-driven sovereign portability, and a supervised participatory dialogue surface — running in production on EU-sovereign and New Zealand-sovereign infrastructure. Discharges the Te Tiriti governance duties Dr Taiuru’s Kaupapa Māori AI Framework asserts; engages the wider personhood inquiry he poses in the companion paper (Paper B — synopsis published).
Distributive Equity Through Structure: A Community-Scale Worked Example of Values Stickiness
A documentary case study offered to the legal-academic research programme on ecosystem power. Documents Village’s Tractatus-framework constitutional architecture as an enactment of values stickiness, grounded in Wittgenstein, Berlin, Ostrom, Alexander, and Te Ao Māori frameworks of indigenous data sovereignty.
Five languages (EN/DE/FR/NL/MI) · CC BY 4.0 · ORCID 0009-0005-2933-7170 →Situated Language Layers for Minority-Language and Indigenous Communities (Paper B Synopsis)
The empirical companion to Paper A. A per-tenant situated language layer trained on the tenant’s own corpus, governed by the tenant’s own authority, and operated on infrastructure inside the tenant’s jurisdictional reach. Five training-discipline rules empirically derived from the project’s thirteen cohorts; nine weight-modification ablation experiments motivating a strict no-weight-modification stance; five Tier-1 cohorts in production; CPU-fallback inference architecture keeping the runtime path entirely outside US-controlled infrastructure. This is the 2-page synopsis — the full empirical paper is deferred to a separate session with verified training-run data, per-cohort evaluation tables, and ablation-result detail.
Why your community's records need to belong to you
A non-technical companion to Paper A v4. The records your community keeps on a SaaS platform — minutes, member rolls, decisions, photographs — are records that platform may change, lose, or hand over without your consent. This work answers the structural problem architecturally, by giving your records cryptographic properties that survive the operator. Written for parish treasurers, sectoral-association secretaries, family-history curators, and anyone running a community organisation on infrastructure they don’t control.
Read the blog post →Why a community needs its own trained AI, not a borrowed one — the why behind Paper B
A non-technical companion to Paper B. Paper A solved data sovereignty; Paper B is about the AI part. Why a community using a frontier vendor model imports the vendor’s authority along with the answers — and what to do instead. Five training-discipline rules learned the hard way; five tenant-type cohorts deployed today; CPU-fallback inference architecture on NZ-sovereign and EU-sovereign infrastructure.
Read the blog post →Sovereign AI Governance at Community Scale: An EU Policy Brief
A policy-audience derivative of the Distributive Equity whitepaper. Three mechanisms — Situated Language Layer, Guardian Agents, Federation — mapped onto the AI Act, EMFA, GDPR Article 9, DSA, and CLOUD Act. Structural audit criteria an adopting community or business can run for itself.
English & German · CC BY 4.0 · derived from the Distributive Equity whitepaper →AI Governance for Communities — A Smaller Room Than You Think
Who decides what AI becomes, and by what authority? Philosophy, philanthropy, sovereignty, and practical governance — including Māori frameworks for data sovereignty and what action ordinary readers can take.
Five articles · CC BY 4.0 · mysovereignty.digital →Why a sovereign record architecture — the why behind the paper
A non-technical companion to Paper A, written for parish treasurers, rūnanga secretaries, family-history curators, and anyone responsible for community records on a US-owned platform. The platform exists, it runs on EU- and NZ-sovereign infrastructure today, and this page explains why it had to be built.
Read the blog post →Mythos and the Economics of Cyberattack
Anthropic’s Mythos model changes the economics of cyberattack permanently. We analyse the three real dangers and what they mean for sovereign infrastructure.
Physical Tenant Isolation: Sovereign Database
No competitor in the NZ small business market offers physically isolated databases on sovereign infrastructure. Our research findings and the architecture we built.
Encryption at Rest: Both Servers Complete
AES-256-CBC encryption at rest now active on both production servers (EU and NZ). All known dependency vulnerabilities remediated. 48-hour patch cycle adopted.