The next generation of geopolitical risk inference

Geopolitics Axiom-0 established inference-native geopolitical risk intelligence—moving beyond static country scores and episodic expert narratives toward continuous, asset-level assessments grounded in proprietary geopolitical, economic, security, and geospatial data.

Geopolitics Axiom-1.0 extends that foundation with a control layer designed for institutional analysts, risk officers, and compliance teams who need transparent, auditable geopolitical intelligence at scale. Analysts can now configure how the model reasons, which sources it draws on, and how outputs appear in reports and analytics—without sacrificing the precision that made Axiom-0 deployable across banks, insurers, and asset managers.

Explainability controls

Axiom-1.0 introduces adjustable explainability settings so analysts can increase the depth of reasoning traces exposed in each assessment. Controls include expanded citation chains, step-by-step inference summaries, confidence decomposition by risk channel, and structured rationales suitable for committee review or regulatory documentation.

Teams can dial explainability up for high-stakes decisions—sanctions exposure reviews, cross-border lending, portfolio rebalancing—or streamline outputs for monitoring dashboards. Every setting preserves Axiom-1.0's inference-native architecture: explanations reflect actual model reasoning paths, not post-hoc summaries bolted onto black-box scores.

Source governance: switch sources on and off

Analysts can now enable or disable source families within Axiom-1.0 workflows—government and policy feeds, conflict datasets, trade and sanctions registries, macroeconomic indicators, geospatial intelligence, satellite-derived signals, and alternative data channels. Source weighting presets allow teams to emphasize official disclosures, open-source intelligence, or proprietary Earthian feeds depending on the use case.

Each assessment carries full source provenance: which inputs were active, when they were last refreshed, and how they contributed to the final inference. This gives risk and research teams the same granular control over geopolitical inputs that underwriting teams expect from climate and NatCat models—making Axiom-1.0 suitable for governed institutional workflows rather than one-size-fits-all reporting.

Propaganda-linked source identification and filtering

Axiom-1.0 can identify sources and narrative channels with known propaganda linkage—state-aligned disinformation networks, coordinated influence operations, and outlets flagged through Earthian's geopolitical source graph. Analysts can switch propaganda-linked sources off entirely, reducing their weight, or quarantine them from final reports and analytics while retaining them for red-team review.

The "Switch off Propaganda" control removes identified propaganda-linked inputs from production outputs by default, producing cleaner geopolitical assessments for investment committees, underwriting panels, and executive briefings. Teams that need full-spectrum monitoring can re-enable filtered sources for scenario analysis or adversarial testing—without contaminating primary decision workflows.

This capability builds directly on Earthian's work with Geopolitics Axiom-0 and the No Propaganda Geopolitics feature: Axiom-1.0 makes propaganda awareness a first-class control, not a post-processing step, so institutions can defend analytical integrity on both sides of contested geopolitical narratives.

Controls for institutional deployment

Control presets can be saved at desk, division, or enterprise level—aligning geopolitical intelligence with internal risk policy, jurisdictional requirements, and client mandates. Audit logs record every control change and source configuration, supporting model governance frameworks and third-party review.

Geopolitics Axiom-1.0 is available through Earthian Hub alongside Axiom-0, Lucid Climate-0, NatCat Lighthouse-0, and other specialized risk SLMs—enabling multi-dimensional risk intelligence where geopolitical controls integrate with climate, credit, and technology inference in a single governed output.