Detection rate
TBD
Publishes at launch
We publish our numbers.
Detection and false-alarm rates, measured on apps with known vulnerabilities. Misses included.
Apps where the answers are already known.
Intentionally vulnerable and previously disclosed apps, so every hit and every miss can be checked.
- Android, iOS, Windows and firmware targets
- Intentionally vulnerable training apps
- Apps with previously disclosed vulnerabilities
- Ground truth fixed before each run
Three numbers, held back until launch.
False-alarm rate
TBD
Publishes at launch
Median time to report
TBD
Publishes at launch
How every run is scored.
01
Freeze the ground truth
Every known vulnerability in each app is listed and locked before the run starts.
02
Run blind
The agents get the binary and nothing else: no hints, no answer key, no retries.
03
A human scores each finding
A reviewer marks every finding as a true hit or a false alarm against the frozen list.
04
Publish everything
Raw results go out as they are, misses included.
Results by platform.
| Platform | Apps | Detection | False alarms | Median time |
|---|---|---|---|---|
| Android | Publishes at launch | Publishes at launch | Publishes at launch | Publishes at launch |
| iOS | Publishes at launch | Publishes at launch | Publishes at launch | Publishes at launch |
| Windows | Publishes at launch | Publishes at launch | Publishes at launch | Publishes at launch |
| Firmware | Publishes at launch | Publishes at launch | Publishes at launch | Publishes at launch |
Note
No cherry-picking: misses are published alongside hits.