The goal of this project is to provide tools and results of passive and active fingerprinting of Machine Learning and Artificial Intelligence Frameworks and Applications using a common Threat Intelligence approach and to answer the following questions:
- How to detect AI/ML backed systems in the Internet and Enterprise network?
- Is AI/ML apps secure at Internet scale?
- What is AI/ML apps security level in a general sense at the present time?
- How long does it take to patch vulnerabilities, apply security updates to the ML/AI backed systems deployed on the Internet?
Number of AI system exposure is growing, America is ahead.
Machine learning frameworks such as Kubeflow, MXNET, Tensorflow, NVIDIA DIGITS are everywhere.
Source (csv, json, images)
Sergey Gordeychik, Denis Kolegov, Antony Nikolaev, "Measuring Artificial Intelligence and Machine Learning Implementation Security on the Internet"