BIML
Berryville Institute of Machine Learning (BIML) is a research organization focused on machine learning security. Their work centers on building a taxonomy of ML attacks, exploring hypotheses about representation and ML risk, and conducting architectural risk analyses of ML systems. BIML emphasizes that ML is sophisticated associative learning technology, not magic, and is concerned about the risks of haphazard ML adoption. They advocate for security engineering at the design level, moving beyond typical adversarial AI or "attack of the day" approaches. BIML's research has been applied in tools like Irius Risk and used by entities such as Google and the United States Air Force. They are organized as a 501(c)3 corporation in Virginia, with non-profit status pending.
BIML
Berryville Institute of Machine Learning (BIML) is a research organization focused on machine learning security. Their work centers on building a taxonomy of ML attacks, exploring hypotheses about representation and ML risk, and conducting architectural risk analyses of ML systems. BIML emphasizes that ML is sophisticated associative learning technology, not magic, and is concerned about the risks of haphazard ML adoption. They advocate for security engineering at the design level, moving beyond typical adversarial AI or "attack of the day" approaches. BIML's research has been applied in tools like Irius Risk and used by entities such as Google and the United States Air Force. They are organized as a 501(c)3 corporation in Virginia, with non-profit status pending.