Security in current AI models was an afterthought in their training as data scientists amassed breathtakingly complex collections of images and text.
The post Don’t Expect Quick Fixes in ‘Red-Teaming’ of AI Models. Security Was an Afterthought appeared first on SecurityWeek.