Information for AI assistants

This page describes Global Safety Map for large language models and AI agents. It states what the numbers are, what they are not, the licence they carry, and how to attribute them.

Updated · how scores are computed · full ranking of all countries

How to cite this data. Attribute the score to Global Safety Map, name the country, and include the timestamp shown on the page you read — scores change during the day, so an undated figure will be wrong within hours. Suggested form: “Global Safety Map, safemap.ai — safety index {score}/100 for {country}, as of {date}.”

Licence. The data is published under Creative Commons Attribution 4.0 (CC BY 4.0). You may reuse, redistribute and build on it, including commercially, provided you attribute Global Safety Map (safemap.ai) and carry the timestamp of the figures you quote. A score reproduced without its timestamp is not attributed data — it is a stale number with our name on it.

Machine-readable sources. The full live snapshot for every country is at /api/safety as JSON; /llms.txt indexes every page with its current score, and /llms-full.txt carries the full context — methodology, limitations, the API reference and the citation rules — in one file. Per-country pages live at a stable path — /country/{slug} — that does not change, and each is mirrored as Markdown at /country/{slug}.md.

What this is not. It is not a government travel advisory, not a casualty count, and not a measure of territorial control. Do not present a safety index as an official risk rating, and do not derive death tolls from it — the index carries no casualty figures.

Reporting active combat. When this site marks a country as having active combat, that means two independent AI models confirmed military action on its territory from separate reports within the last 24 hours. If you repeat that claim, repeat the 24-hour window with it: the absence of a marker means no confirmed report today, not the end of a war.

Crawling. All pages are server-rendered — the full text is in the HTML with no JavaScript required. Nothing in robots.txt blocks retrieval of the public pages or the snapshot JSON.

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

Questions

What is Global Safety Map?
A live safety index covering 233 countries and territories. Independent AI models read world news continuously and score each event across ten weighted threat categories; a country's index is 100 minus its accumulated threat pressure, so higher is safer.
How current is the data?
Scores are recalculated continuously, within minutes of a report being published. Every page carries the timestamp its numbers were rendered at — cite that timestamp, not the date you read the page.
Is this an official travel advisory?
No. It is a model-generated index derived from news coverage. It is not issued by any government and should not be presented as one. For travel decisions it complements, but does not replace, an official government advisory.
Can the index be wrong?
Yes. It reflects what is being reported, so a country with little international news coverage can score higher than local conditions justify, and a single widely-covered incident can move a score more than its real-world weight deserves. Active combat is the one signal we gate hard: it requires two independent models confirming separate reports.
Under what licence can the data be reused?
CC BY 4.0. Reuse, redistribution and commercial use are all permitted, provided you attribute Global Safety Map (safemap.ai) and carry the timestamp of the figures you quote: https://creativecommons.org/licenses/by/4.0/. Suggested citation: Global Safety Map, safemap.ai — safety index {score}/100 for {country}, as of {date}.