How the Safety Index works

Every country's score is computed from real news, continuously. No editors, no opinions in the loop — a fixed formula over machine-assessed events. Here is the entire pipeline and the exact math.

Questions, partnerships or data-removal requests: contact@safemap.ai

The 0–100 scale

Higher is safer. The scale is divided into six verdict bands — the same colors used on the live map and on every country page:

The pipeline

1. ParseWe continuously collect news from 70+ leading media outlets across the world, around the clock. Duplicates are filtered before analysis.
2. AssessEach news item is assessed by large language models acting as threat-intelligence analysts — drawing on the country's already-known context and on knowledge distilled from vast amounts of previously evaluated news.
3. ScoreThe assessed event feeds a deterministic formula (below): it raises that country's threat level in one category. Nine safety categories determine the index; nuclear events feed separate context.
4. Cool downThreat levels continuously cool toward each country's long-term baseline, so elevated scores fade once the news flow subsides — the map reflects what is happening now, not a permanent stain.

The formula

The Safety Index of a country is:

where T_c is the current raw threat level in one of nine safety categories. Their raw caps sum to 88 and are proportionally normalized to 100, so S always lands between 0 (every safety factor saturated at once) and 100 (no active pressure in the nine safety factors).

Each assessed civilian event moves one category:

• σ — severity 0–10, assigned by the model;
• m — involvement multiplier: 1 for ordinary events. Military events use one of the four channels below; disputed territory controlled by the country uses 0.7 inside the operations-abroad channel;
• k — single-event impact: 0.15 for military and nuclear events, 0.25 for every civilian category. Nuclear events update only the Nuclear index and do not move the Safety Index;
• W_c — the raw category cap; the table below shows its normalized safety contribution.

Four kinds of military exposure

Independent raw limits are 30 for sustained ground combat, 24 for sustained air campaigns, 12 for localized strikes and 6 for operations abroad. Their normalized limits are 34.1, 27.3, 13.6 and 6.8 safety points. Reporting volume cannot promote one kind to another.

Calculate each model first, then average model results. Each channel cools toward its reviewed seed floor. These limits are methodological choices, not measured probabilities.

Unclassified legacy activity receives no new events and cools toward an explicitly preserved previous seed minimum. Once reviewed channel floors replace that minimum, the remaining unclassified value cools toward zero. During the transition military is max(legacy, M). Original history remains archived; retrospective estimates carry separate review evidence.

Category weights (W_c)

Σ W_c = 100

Nuclear threats & incidents

Our Nuclear index (0–100) tracks nuclear weapons development, tests, radiation incidents and threats to use nuclear weapons. It includes both countries making threats and explicitly threatened countries. The underlying records are preserved. The index has zero effect on the Safety Index and is not a probability of nuclear war. The separate “Has nuclear weapons” status comes from SIPRI; SIPRI does not provide this index.

Consensus of independent models

Several LLMs assess the same news stream independently, each maintaining its own set of threat levels. The default "Global Consensus" view is the per-category mean:

One model overreacting to a headline gets averaged down by the others. You can switch the map to any single model's lens in the header to see how assessments differ.

Cooling down

Every 5 minutes each category sheds 0.2% of its elevation above the country's long-term baseline — a half-life of roughly 1.2 days. The score is an equilibrium between event inflow and this steady cooling: sustained war coverage keeps a country pinned near its maximum penalties, while a country that merely generates a lot of routine news settles mid-scale instead of sinking to the same floor as an actual war zone. The baseline itself encodes structural, slow-changing risk (long-running conflicts, chronic instability), so a country in an active war does not cool to "safe" during a quiet news day.

Freshness & stability

Scores are recalculated continuously, not on a daily batch. Every country page shows its exact last-updated time and a 30-day trend. A country whose score is not moving carries a “Stable for N days” badge: N counts consecutive days in which the day-over-day score change stayed below 0.5 points. The badge appears from 7 days; a country with confirmed active combat shows a combat marker in the same place instead.

Combat confirmation

The combat marker is the one signal we refuse to take from a single voice. It appears only when two independent AI models, reading separate news reports, both confirm military action on a country's territory within the last 24 hours. When the confirmations stop, the marker goes — its absence means no confirmed report today, not a declared peace.

Boundaries & disclaimer

Map geometry comes from the public-domain Natural Earth dataset (via world-atlas), which draws state boundaries by de facto control rather than legal claim. The boundaries, names and designations shown on this map are carried over from that open dataset as-is. They do not imply the expression of any opinion whatsoever on the part of the project's authors concerning the legal status of any country or territory, or concerning the delimitation of its frontiers or boundaries.

Hatched areas on the map mark territories from Natural Earth's breakaway and disputed-areas layer (Crimea, Kashmir, Western Sahara and others). They are shown for context only: a disputed territory carries no safety index of its own, and selecting one deliberately opens no country. An event located inside such a territory moves the index of the state that de facto controls the ground — at 70% of the operations-abroad channel impact — not the index of the state that holds the claim under UN-recognized borders.

Combat markers, scores and events are generated automatically from publicly available news reporting by AI models, without human verification. They may be wrong, incomplete, mislocated or outdated, and are not statements of fact about any location; scores react to reporting volume as well as events themselves.

Everything on this site is provided "as is", for informational purposes only, without warranty of any kind. The authors accept no liability for any decisions made on the basis of this map. This is not travel, security or legal advice.

See it applied

Is United States safe? Is Japan safe? Is Mexico safe? Is Thailand safe? Is Ukraine safe?

Open the live map →