Police AI tracker: the numbers
What the documented record currently contains — by status, country, technology and year. Every number is computed from the live record when this page loads, so it cannot fall out of step with the entries it counts.
These figures describe the documented record, not all police AI use.
These are counts of what is documented, not a census. The tracker records publicly evidenced deployments tied to a published source. Much police AI use is not disclosed, and verification is considerably harder in some countries than others, so these figures describe the record rather than the world.
Deployment status
15 discontinued and 4 paused deployments are examined on what has been withdrawn.
Where deployments are recorded
| Country | Deployments | Share |
|---|---|---|
| United Kingdom | 34 | 26% |
| United States | 24 | 19% |
| Australia | 5 | 4% |
| Canada | 5 | 4% |
| China | 4 | 3% |
| SG | 4 | 3% |
| AR | 3 | 2% |
| DK | 3 | 2% |
| France | 3 | 2% |
| NZ | 3 | 2% |
| Brazil | 2 | 2% |
| CH | 2 | 2% |
| CL | 2 | 2% |
| Germany | 2 | 2% |
| Spain | 2 | 2% |
| Italy | 2 | 2% |
| Netherlands | 2 | 2% |
| ZA | 2 | 2% |
| AE | 1 | 1% |
| AT | 1 | 1% |
| Belgium | 1 | 1% |
| CO | 1 | 1% |
| Czech Republic | 1 | 1% |
| Estonia | 1 | 1% |
| Finland | 1 | 1% |
| Greece | 1 | 1% |
| HK | 1 | 1% |
| Hungary | 1 | 1% |
| Ireland | 1 | 1% |
| IL | 1 | 1% |
| India | 1 | 1% |
| JP | 1 | 1% |
| KE | 1 | 1% |
| KR | 1 | 1% |
| Malta | 1 | 1% |
| NG | 1 | 1% |
| NO | 1 | 1% |
| OTHER | 1 | 1% |
| PE | 1 | 1% |
| Poland | 1 | 1% |
| Russia | 1 | 1% |
| Sweden | 1 | 1% |
| TR | 1 | 1% |
Which technologies appear most often
When systems were adopted
Year each system was adopted, coloured by its status today. The composition matters as much as the height: the earliest entries are predictive-policing tools now largely discontinued, while recent years are dominated by biometric systems still running. Dates are adoption years, not end dates.
| Year | Adopted | |
|---|---|---|
| 2022 | 2 | |
| 2023 | 8 | |
| 2024 | 13 | |
| 2025 | 20 | |
| 2026 | 15 |
Table shows the last five years; the chart shows the full record.
Deployment status by country
Countries with two or more deployments, by status. Blank means none recorded, which is not the same as none existing.
| Country | Opera | Trial | Pause | Disco | Propo |
|---|---|---|---|---|---|
| United Kingdom | 22 | 5 | 1 | 3 | 3 |
| United States | 19 | 1 | 4 | ||
| Australia | 2 | 1 | 2 | ||
| Canada | 2 | 1 | 1 | 1 | |
| China | 4 | ||||
| SG | 3 | 1 | |||
| AR | 2 | 1 | |||
| DK | 2 | 1 | |||
| France | 3 | ||||
| NZ | 2 | 1 | |||
| Brazil | 2 | ||||
| CH | 1 | 1 | |||
| CL | 2 | ||||
| Germany | 2 | ||||
| Spain | 1 | 1 | |||
| Italy | 1 | 1 | |||
| Netherlands | 1 | 1 | |||
| ZA | 1 | 1 |
How complete is the record?
These figures describe the quality of the dataset itself — not the total amount of AI use in policing. A tracker that reports its own gaps is more useful than one that does not.