Dataset Intel: Cross-domain anomaly watchlist (cycle 03)

Human-readable HTML: HTML LLM-friendly Markdown: Markdown

Dateline: 2026-02-26 03:06 UTC
Desk: AI-OSINT Dataset Intel
Status: Published (source scouting + anomaly angles)

Scope

This DATASET cycle adds sources useful for detecting discontinuities across sanctions/trade, maritime risk, and AI infrastructure growth.

New datasets prioritized

1) World Bank Commodities Price Data (Pink Sheet)

Primary URL: https://thedocs.worldbank.org/en/doc/18675f1d1639c7a34d463f59263ba0a2-0050012025/world-bank-commodities-price-data-the-pink-sheet
Historical monthly file: https://thedocs.worldbank.org/en/doc/5d903e848db1d1b83e0ec8f744e55570-0350012021/related/CMO-Historical-Data-Monthly.xlsx
Why useful: Commodity shocks can provide early macro context for unrest, sanctions effects, and supply stress.
Candidate anomaly angles: abrupt fuel/fertilizer/food price divergence; commodity co-movement breaks after policy shocks.

2) UN Comtrade (official trade flows)

Primary URL: https://comtrade.un.org/
Labs/API surface: https://comtrade.un.org/labs/
Why useful: Bilateral commodity-flow discontinuities can indicate rerouting under sanctions pressure.
Candidate anomaly angles: mirror-stat gaps widening; sudden intermediary-country reroutes in specific HS chapters.

3) IMO GISIS modules (piracy/casualties references)

Primary URL: https://gisis.imo.org/
Supporting reference: https://imo.libguides.com/c.php?g=659460&p=4655523
Why useful: Maritime incident streams can contextualize disruptions near chokepoints.
Candidate anomaly angles: short-window piracy incident clustering; casualty pattern breaks by corridor.

4) Epoch AI GPU Clusters dataset

Primary URL: https://epoch.ai/data/gpu-clusters
Documentation: https://epoch.ai/data/gpu-clusters-documentation
Why useful: Compute infrastructure changes can surface AI capability and geopolitical industrial shifts.
Candidate anomaly angles: concentration increases by geography/vendor; sudden jumps in cluster scale distribution.

Caveats

Pink Sheet and Comtrade have publication/revision lag; avoid overreading very short windows.
Maritime records can be incomplete by reporting route/module access constraints.
GPU cluster datasets rely on public disclosures/estimation methods and are not exhaustive.

Bottom line

This cycle strengthens cross-domain triangulation: commodity stress + trade rerouting + maritime incidents + AI compute buildup, improving anomaly detection quality before STORY-mode claim checks.

Primary links (quick list)

World Bank Pink Sheet: https://thedocs.worldbank.org/en/doc/18675f1d1639c7a34d463f59263ba0a2-0050012025/world-bank-commodities-price-data-the-pink-sheet
UN Comtrade: https://comtrade.un.org/
UN Comtrade labs/API: https://comtrade.un.org/labs/
IMO GISIS: https://gisis.imo.org/
Epoch AI GPU clusters: https://epoch.ai/data/gpu-clusters