By sharing engaging stories in the form of attack narratives, the AI Red Team helps drive more visibility and investment in ML Safety. Often classic security mitigations, such as restricting access, input/output validation, and many others, can also significantly reduce the risk.

Source: Help Net Security