Risk
Absolute counts say little about risk. Measured against population, the picture changes: the countries with the most cases are not the countries where an individual is most exposed.
Attacks per million adults
Two adjustments make the comparison meaningful. Only attacks aimed at personal assets are counted — raids on mining farms, crypto ATMs and company treasuries are a business risk, not a holder's risk, and account for 62 of the 670 cases. And the window is limited to the last 36 months, because the phenomenon changed scale after 2023.
The denominator is the population aged 15 and over rather than the total population, so that countries with young populations are not penalised: Nigeria is 59% adult, Hong Kong 89%.
Taiwan and France lead at levels that cannot be told apart statistically — their confidence intervals overlap.
Confidence intervals
The horizontal line through each bar is a 95% Poisson interval. 19 of the 22 countries shown have fewer than thirty cases in the window, so their intervals are wide, and neighbouring countries in the ranking are usually indistinguishable.
The number in brackets is the raw case count behind the rate.
Frequency against severity
A country can rank low on frequency and high on harm. In Iran and Indonesia every recorded case ended with an injured or killed victim; in Brazil, 29% did.
Attacks per crypto-asset holder
The denominator that would matter most is the number of people who actually hold crypto-assets. It is measured by public survey in only a handful of countries, and those are the only ones shown here.
On this basis a holder in France is attacked roughly 11 times as often as a holder in the United States, and 6 times as often as one in the United Kingdom — a gap too large to be explained by differences in press coverage between these countries alone.
For every other country, no reliable ownership estimate exists. The figures most widely quoted in the press come from commercial operators that publish no methodology, and diverge by more than a factor of twenty between sources for the same country. No per-holder rate is calculated beyond this table.
Survey quality
The Federal Reserve figure comes from an address-based probability sample. The Financial Conduct Authority and Bank of Canada use quota-based online panels. The Adan/Ipsos figures are from an online panel commissioned by an industry association, and should be read as the weakest of the set.
Ownership rates are measured on adults aged 18 and over and applied here to the population aged 15 and over, which slightly overstates the number of holders and therefore understates the rate.
| Country | Attacks per 100,000 holders per year | 95% interval | Ownership rate | Source of the ownership rate |
|---|---|---|---|---|
| France81 cases | 0.43 | 0.34 – 0.53 | 11.0% | Adan / Ipsos, 2026 Barometerheld crypto-assets at the time of the survey |
| Belgium5 cases | 0.11 | 0.04 – 0.26 | 15.0% | Adan / Ipsos, 2026 Barometerheld crypto-assets at the time of the survey |
| United Kingdom10 cases | 0.07 | 0.03 – 0.13 | 8.0% | FCA, Cryptoasset Consumer Research 2025 (wave 6)held crypto-assets at the time of the survey |
| Canada7 cases | 0.07 | 0.03 – 0.14 | 9.9% | Bank of Canada, Bitcoin Omnibus Survey 2023held BITCOIN only (Ether 5%, others 3%) |
| Netherlands5 cases | 0.05 | 0.02 – 0.13 | 20.0% | Adan / Ipsos, 2026 Barometerheld crypto-assets at the time of the survey |
| United States32 cases | 0.04 | 0.03 – 0.06 | 9.0% | Federal Reserve, SHED 2025 (table 33)bought or held crypto as an investment in the previous 12 months |
| Germany5 cases | 0.03 | 0.01 – 0.08 | 7.0% | Bundesbank, Payment Behaviour in Germany 2025has already bought crypto tokens or paid with them — a cumulative measure, not holdings at a point in time |
| Italy2 cases | 0.01 | 0.00 – 0.04 | 13.0% | Adan / Ipsos, 2026 Barometerheld crypto-assets at the time of the survey |
Limits of these rates
Every figure on this page counts attacks that were reported and found, not attacks that happened. Countries where local crime reporting is dense and prosecutors comment publicly will rank higher than countries where the same events go unpublished, and the gap measures visibility as much as violence.
The dataset records the Reporters Without Borders press freedom score for each country so that this can be checked. Where it falls below 40 out of 100 — Iran at 17, Russia at 23, Türkiye at 28, India at 32, Hong Kong at 39 — the rate shown should be read as a floor.