Edition and evidence classification
Every published material claim should say what was tested, where, which exact version was used, which independent party (if any) reviewed it, what failed, and which release gate remains open. The evidence table below is deliberately conservative; it is a status summary, not a newly conducted software test.
| System | Existing public claims | Interpretable status | Remaining acceptance records |
|---|---|---|---|
| Exchange backend | Local backend order/wallet/ledger tests reported in October 2026 | Qualified only within named local scope; do not claim live regulated venue | Signed native device qualification, independent CI, issuer/custody controls, legal authorization |
| NIKI | Bounded WASM sensor repair 65,536/65,536; Granby candidate had 256 wrong cases | Narrow pass and separate failed research retained; not general autonomous invention | Independent unfamiliar-task suite, reproducible original NIKI runtime, change/repair receipts |
| FarmNodes | Logical type/binding registries and single-machine farm workers demonstrated | Three independent physical hosts not verified at October gate | Enrollment/identity/isolation/reproduction on 3 actual independently qualified machines |
| ADAM / ENTITY | Local grants/signatures and revocation-related evidence recorded | Production owner-signed authority for complete system pending | Real scoped production keys/grants, distributed revocation and negative security tests |
| BTDU | Exact-byte replay and local integration evidence recorded | Not proof of superior LLM-model capacity, compression or all workloads | Independent representative reconstruction, quality, speed, memory and storage benchmarks |
| Bridge | Synthetic signed/encrypted intake and limited local tests | Real accounts and third-party connectors/production isolation not certified | Approved providers, user journeys, deletion/revocation and external security assessment |
| Circuits / Composer | Typed host tests, selected callable conversions and partial native app construction | Arbitrary conversion/parity and full production GUI not established | Frozen runnable parity benchmark; dependency/rights tests and accessibility checks |
| Investment issuer | Conceptual robotics DCOs and U/A/I structures documented | No actual investment issuance or real client-funds settlement evidenced | Legal issuer, full offering/disclosure, qualifications, custody/banking, regulatory advice |
| Boundary Tech Generation | 36-page proposed community workshop blueprint dated October 22 | Proposal only; no registrations/venue/volunteers or sessions represented as approved | Host agreement, screening, equipment, insurance, guardian consent, tested lessons |
Evidence receipt schema for future publication
| Field | Expected content |
|---|---|
| Claim identifier | Human-readable label and linked published version |
| Frozen artifact | Exact version, hash and distribution format |
| Test environment | OS, hardware, runtime, settings, dependencies, locality |
| Rights and authority | Approving owner and permitted test scope |
| Test design | Predeclared criteria, normal/negative/holdout cases |
| Result | Observed pass and fail details, retained counterexamples |
| Independent reviewer | Organization, identity, separation and scope where applicable |
| Limitations | What the result does not demonstrate |
| Revision and correction | Time, reason, remediation, link to superseding artifact |
Publication correction policy
Superseded numbers or test descriptions should be labelled as historic, never quietly reused as current. Private source can remain private while public claims are individually checkable through signed, redacted receipts or controlled evaluations. The absence of a publicly inspectable receipt means the public description should be treated as a claim, not a verified independent proof.
AI Creation Council and FarmNode research — audited update
Audit figures: council software 40/40 local tests; FarmNode controller 70/70 across six suites, partly overlapping earlier council tests; two genuine local Qwen3-8B research methods fully passed their 65,536-case bounded WASM suites. A deliberately incorrect sample and runaway loop were rejected. The Granby model candidate failed 256/65,536; a deterministic human-reviewed correction subsequently passed 65,536/65,536 in isolation.
Limits: 0 independently qualified physical FarmNodes; 0 owner-signed production admissions; 0 proven real hosted independent multi-model council comparisons; no cross-domain 25-task qualification and no general invention or commercial value claim. Original internal audit receipts exist but are not independently publicly certified.
Read the complete AI Council research guide → · Public operator manual → · Printable evidence report →
UPDATED AUDIT · 23 OCT 2026 · 18:26 UTC
Latest Council and FarmNode qualification evidence
The completed 18:26 UTC audit reports 114/114 local tests in nine suites, including the earlier 70 tests (not 184 independent results). The 34 physical-host security tests qualify policy/refusal contracts only, not three real computers. Four explicitly approved hosted routes were zero-priced at the 18:17 UTC catalogue snapshot; no authenticated hosted inference comparison was completed.
Research controller: 5,125 registered types; 2,305 registration-only bindings; 16 source-backed local methods; six original messages consolidated to four unique research requirements. Three historical source-backed evaluation packs now exist for forestry (8 holdouts), agriculture (8), and emergency-response metadata (9). These 25 recorded cases are not 25 new algorithm inventions. Forestry median-change beat naive forecasting on the recorded sample, while agriculture median-change performed worse; emergency-response checks verified historical metadata/timestamps only, not event truth or live warning decisions.
The original Granby model-authored algorithm failed 256 of 65,536 cases. A separately documented engineer-reviewed deterministic repair passed 65,536/65,536 in isolated WebAssembly. This is not NIKI self-repair, a domain-qualified commercial invention, or owner-granted Genome admission. Two other different genuine local Qwen3-8B small-domain candidates passed bounded 65,536-case tests apiece.
Release remains HOLD: 0 of 3 independent physical FarmNodes qualified; no real hosted cross-model algorithm-comparison evidence; the separate 25-task invention benchmark is incomplete; no production ENTITY/Genome owner grant; and no rights or authorization for automatic Exchange investment issuance.
Full Council qualification report → · Download qualification PDF → · Three sector evaluator packs → · Download evaluation PDF →