Cyber Grand Challenge Darpa Machine Automated Cyber Attack

The DARPA competition utilizes machine learning to generate software exploits in assembly and Python using Neural Machine Translation for natural language descriptions. This includes modifying Java bytecode and filtering HTML requests.

---

ctfwiki’s intro on CGC

analyze source code first, then plan attack or fix code

cgc’s github repo and website

search for darpa cgc on github

cyber-challenge Some toy examples, to demonstrate ideas that could be used in DARPA’s Cyber Grand Challenge including modifying java bytecode and filter out html requests on the fly

EVIL (Exploiting software VIa natural Language) is an approach to automatically generate software exploits in assembly/Python language from descriptions in natural language. The approach leverages Neural Machine Translation (NMT) techniques and a dataset that we developed for this work.

Topics

linux exploit encoder assembly decoder dataset seq2seq shellcode nmt software-exploitation codebert

Resources

Readme

License

GPL-3.0 license

Stars

13 stars

Watchers

3 watching

Forks

1 fork

Releases

No releases published

Packages

No packages published

Contributors 2

@piliguori

piliguori Pietro Liguori

@taisazero

taisazero Erfan Al-Hossami

Languages

Python

97.6%

Shell

2.0%

Other

0.4%