This repository is not an AI Futures Project publication, does not speak for AI 2040, and does not certify a treaty or verification regime. It documents the relationship between public AI 2040 concepts and concrete open-source evidence, data-engineering, and research tasks so other researchers can reproduce, challenge, or extend the work.
Contribute by role
| Role | Concrete entry point |
|---|---|
| Grid / power systems | Add a connection region, reconcile utility/service records, or work the 83 unlocated MISO requests. |
| Remote sensing / geospatial | Resolve the one remaining physical target, improve scene selection, or validate thermal/SAR/optical and footprint linkage. |
| Current unresolved physical target: OpenAI Stargate UAE has an attributed September 11, 2026 Reuters area-level lead near Al Dhafra Air Base, but the retained evidence still lacks a canonical facility parcel or coordinate; the repository therefore keeps the site unresolved rather than fabricating a target. | |
| Data engineering / provenance | Strengthen manifests, SHA-256/byte checks, schemas, generated-artifact validation, and refresh reproducibility. |
| Research / policy | Close site-specific evidence tasks and test the Phase 1 versus Track 3 boundary without collapsing uncertainty. |
| Verification / security | Audit claim gates and connect public evidence to independently testable verification mechanisms. |
Primary AI 2040 references
AI 2040's public Track 3 page identifies the goal as verifying the absence of secret compute. The Covert AI Projects supplement describes accounting for pre-deal compute, tracing sales/resales to physical end use, and locating remaining unaccounted compute.
Observation-layer scope
The interconnection core is North American public-evidence coverage. Its strongest direct contribution is public accounting and observability of large compute/load infrastructure and associated entities; it is not the bilateral U.S.–China verification layer of a mature Track 3 regime. The public-source boundary also means a genuinely covert project could avoid these signals. The defensible capability claim is to raise the cost of hiding and map the ambiguous middle, not to claim discovery of every secret facility.
Plan A → repository contribution map
This table is generated from data/track3/ai2040_plan_a_mapping.json. Status describes this repository's current implementation state, not the truth of the underlying world.
| AI 2040 element | Repository evidence / artifact | Status | Next contribution |
|---|
Large-load verification universe
The contribution now includes a public North American large-load universe, a 93-site Epoch AI connection-research layer, a 93-site public project-enrichment bridge, a complete 93-site remote-sensing target ledger, and a source-backed transformer-event ledger. Project records are deliberately separated from aggregate evidence so overlapping utility, regulator, queue, and planning populations are not double-counted.
Machine-readable research layer
For researchers and tooling, the repository exposes both a Plan A traceability table and an audit-chain schema. These separate AI 2040's proposed verification concepts from the evidence actually ingested by this repository.
Reproducibility and CI audit
The repository now records a complete commit-level GitHub Actions audit in data/ci/commit_audit.json, alongside the human-readable audit. Track 3 data-pipeline success is reported separately from Pages deployment status so a website build cannot be mistaken for a successful evidence rebuild.
Reference the work
AnonRish, Continental Large-Load Interconnection & Telemetry Registry: AI 2040 Plan A Track 3 research contribution (2026). https://github.com/AnonRish/continental-load-registry
For scholarly use, cite the repository and the upstream source/dataset that supports the particular claim. The machine-readable citation metadata is in CITATION.cff, with human-readable guidance in CITATION.md.
What to contribute next
The most consequential extensions are not another dashboard layer. They are evidence chains that connect production → sale → transfer/resale → current owner → physical location → inspection, plus a separately auditable pathway for unrecorded compute → external discovery → candidate site → physical evidence → accounting resolution.
Use the Track 3 evidence issue template or submit a focused pull request with the source, dates, raw value, normalized value, scope, and provenance.