What the catalogue shows
Every chart below is computed live from the 276 entries in the database, with percentages taken over the 271 that are incidents or campaigns rather than agency benchmarks. These are patterns in what has been publicly reported and catalogued here, which is not the same as patterns in what happened. Coverage is denser for recent years and for English-language reporting.
Entries per year, and how many involved AI
The rise after 2022 reflects both a real increase in reported social engineering and denser coverage of recent years. The red share is what is new: attacks where a cloned voice or a synthetic face did the persuading.
How the attacker reached a person
Counting primary vectors only.
Sectors targeted
Where the victim organisation sat.
What it led to
Entries can have more than one outcome.
What the money figures measure
Published loss numbers are not interchangeable. Only the first two kinds are summed anywhere on this site.
Evidence quality
How firmly each entry is established, and how many involved AI.
These aggregates are available as JSON at impact.netarx.com/api/stats. For how entries are collected and classified, see the methodology.