Machine Learning (ML) security has a history in security lasting 20 years

Machine Learning (ML) security dates back to 2004, with several experienced teams managing AI risks for over a decade. Despite the emergence of new 'AI red teamers,' there is confusion regarding their roles, with activities like bias testing and brand safety work being labeled under 'red teaming.' The term 'red teaming' is used loosely, even in events like the Generative Red Team at DEFCON 31

While there are new developments in ML Security, the fundamentals remain constant. The talk at SANS AI in Cybersecurity Summit 2024 by Sven Cattell highlighted the importance of understanding the history and learning to be effective in the field.
https://www.youtube.com/watch?v=0zq75T_RYek