Using AI to discover undisclosed vulnerabilities in open-source - Aikido Intel

We conducted a research project to use AI LLMs to discover how many vulnerabilities in open-source projects are patched, but never disclosed. To do this we trained LLMs to read changelogs in projects and discover when security fixes were applied but no CVE was created yet.
We discovered the 67% of all projects never were publically disclosed including some pretty scary examples.
Checkout Aikido Intel - intel.aikido.dev
Checkout our writeup - aikidosec.com/...

[@coom07]

Also don't forget they have some overhead issues in the usa with the vuln reporting workflow

[@chetangiradkar]

you are doing god's work. +100 karma points to you!

[@pentabular]

How can I do this? Join the team?

[@aikidosecurity]

We are always looking for people to join the team www.aikido.dev/careers

[@coom07]

​@@aikidosecurityDo I have to be Belgium citizen? I would like to apply to the cat 😺 position if necessary :/