Zen and the Art of Adversarial Machine Learning

Explore the security concerns and potential pitfalls of machine learning integration in this 44-minute Black Hat conference talk. Delve into the world of adversarial machine learning, examining issues such as PII extraction from language models, theft and bypassing of classification models, and biased decision-making in various industries. Learn about operational guidance, adversarial ML techniques, and capability development to address these challenges. Gain insights into topics like lossy compression, distance metrics, OSINT, inference traffic, and common files used in ML security. Discover essential tooling and best practices to mitigate risks associated with the rapid adoption of machine learning across diverse business processes.