Tyche Institute · AI trust-stack evidence ecosystem

Where AI becomes visible — and where it does not yet.

Tyche builds observatories, corpora, and claim gates that map where AI systems, capital, compute, patents, standards, public institutions, and legal decisions become visible in public evidence — and which gates must close before stronger ownership, compliance, deployment, or crisis claims become safe. We link partial evidence surfaces without fusing them into one master claim.

Observatories17connected evidence surfaces across six contours
Atlas canonical records162,994deduplicated public-metadata records linking law, standards, patents, and cases
Patent canonical records151,890from a repair-first, reproducible patent-evidence pipeline
Open data deposits2archived on Zenodo with DOIs (patent pipeline; standards-fit package)
Tyche AI Trust Stack Ecosystem Atlas connects public evidence, private core, patents, standards, legal-temporal rows, and factory gates Interactive — click any highlighted observatory to open it ↗ Atlas / Claim-Support Corpus source families, law/standards-to-evidence patent/literature traces, claim ladders Connective role: evidence objects, not one fused dataset denominator + source family + proof ceiling stay together Public AI Visibility Public AI Registry + Procurement Gap GAIA + PALLAS + LIMEN Scope: disclosure chain, not public-sector limit Private Core And Patents AION counterparty / filing routes Patent Topology pipeline Trust Turning Point theory Hook: who owns the stack? Answer gated Standards And Evidence Carriers SIGIL + EUDIW Race Computational Trust + VESTA RHR procurement / PQC / PKI signals Standards structure evidence; they do not certify Legal-Temporal Accountability LIMEN + Temporal Commons NOMAD normative diffs CLIO legal-decision support Rows become useful before they become comparable Evidence Infrastructure Core Freshness/replay, PKI governance, verification Public evidence registry, patent cartography Runnable / inspectable fixtures with claim ceilings Research Factory Control Harvest-mode governance for the portfolio Readiness scoring, package gates, review passes Operational detail kept internal, not disclosed Shared grammar: owner surface -> denominator -> source family -> proof ceiling -> blocked reading -> package/checksum route No single layer proves deployment, ownership, compliance, safety, market control, legal truth, or official assurance.
Six contours

One ecosystem, six surfaces of visibility

1

Public AI visibility

Procurement traces, registries, named systems, country-route evidence, and edge-case accountability — a chain of partial surfaces, not one register.

2

Private capital & patents

Public-equity behaviour, counterparty signals, and a reproducible patent-evidence pipeline map the private core's visibility channels before ownership.

3

Standards & carriers

Standards fragments, wallet and credential specs, receipts, and typed fixtures structure evidence — but do not certify compliance by themselves.

4

Legal-temporal accountability

Edge cases, regulatory-lag timing, normative diffs, and legal-decision claim support — evidence that is useful before it is comparable.

5

Research-factory control

Readiness scoring and package gates keep every claim to its ceiling. Operational detail stays internal and is not publicly disclosed.

6

Atlas: the connective map

A versioned, metadata-first corpus that joins all five surfaces as evidence objects with shared grammar and explicit claim ladders.

Observatories

Explore the surfaces

Patent Topology

Repair-first patent-evidence pipeline (647k→152k records).

patent.eatf.eu →

Legal-Temporal (CLIO · NOMAD · Temporal Commons)

Legal decisions, normative diffs, regulatory-lag timing.

legal.eatf.eu →
Scope

This program maps evidence visibility. It does not assert who owns the AI trust stack, ownership concentration or market control, compliance, certification, deployment, public-sector prevalence, safety, or legal truth. Public observatory pages expose statistics, methods, and claim boundaries; raw corpora and operational detail remain controlled. Companion studies and data deposits each carry their own verification ledger and explicit claim ceilings.

© Tyche Institute · Tallinn, Estonia · An AI trust-stack evidence-infrastructure research program. Flagship article · Atlas · Pallas