the pilot: results so far.
what we scanned, what we found, and where. every figure comes from our own dataset. no affected application or organisation is named, and nothing here could identify or exploit an unfixed issue.
from scan to fix.
each step is a share of the one before. most of what the scanner flags never becomes a report: a person confirms every finding first.
- Apps examinedstatic analysis of public APKs
- Findings verified by handcandidates confirmed by a person · 1% of the step before
- Reports sentto the owner or the national CERT · 100% of the step before
- Acknowledgedthe owner or CERT replied · 28% of the step before
- Confirmed fixedre-verified on a new build · 18% of the step before
findings by sector.
the kind of app each finding sat in. public services, banks and health apps are where a leaked secret reaches the most people, so that is where we look first.
install bands.
a flaw in a 100k-install app matters to 100k people. the biggest ones matter to hundreds of millions. each column is an install band; its height is how many of our findings sat there.
- 5 of 368 findings sat in apps with 500M+ installs5500M+
- 8 of 368 findings sat in apps with 100M+ installs8100M+
- 19 of 368 findings sat in apps with 50M+ installs1950M+
- 45 of 368 findings sat in apps with 10M+ installs4510M+
- 86 of 368 findings sat in apps with 1M+ installs861M+
- 75 of 368 findings sat in apps with 100K+ installs75100K+
- 51 of 368 findings sat in apps with 10K+ installs5110K+
- 44 of 368 findings sat in apps with 1K+ installs441K+
- 35 of 368 findings sat in apps with under 1K installs35<1K
where we looked.
findings in 70 countries so far. we start with the apps that matter most to the most people: government, banks, health, transport.
all 70 countries, on the globe →countries with the most findings
how reports travel.
where an affected app belongs to a government or public body, we route through the national cert and let them lead. private owners hear from us directly, through their security contact or bounty program where one exists.
- Through a national CERT151
- Direct to the owner227
Argentina
Austria
Bahrain
Bangladesh
Bhutan
Brazil
Cape Verde
Colombia
Croatia
Côte d'Ivoire
Denmark
Ecuador
Egypt
France
Georgia
Germany
Greece
Hungary
India
Indonesia
Ireland
Italy
Japan
Kenya
Montenegro
Netherlands
New Zealand
Nigeria
Peru
Poland
Romania
Serbia
Singapore
Slovakia
Slovenia
South Korea
Spain
Sri Lanka
Sweden
Taiwan
Turkey
UAE
UK
USA
Uganda
Ukraine
Uruguay
what we publish.
every figure on this page is stripped to sector, country, install band and status: no secret, no package name, no organisation, no host. we never publish a live secret, and we never describe an unfixed issue in a way that could help anyone find it. a fixed, public-interest issue, reported through the national cert, may instead become a named case report; everything else stays anonymous and aggregated, as here.