These are all great topics for not only cybersecurity teams but everyone using or implementing AI solutions. One that stands out to me is “Spotting AI Biases & Hallucinations”. I say this because without enough data or input to go off of the biasing aspects could get you in a lot of trouble like it did Workday. Then again it went a little deeper than that for them so follow the link and read the detail. The other point to make is on the same note as lack of data or reliable authentic and accurate sources AI can and will “fill in the blanks”, “make things up” or add “2+2 to get 5”, this is call hallucinations. So the data is crucial! On the flip side the wrong or company confidential/sensitive data may also be dangerous. So there is a fine line between security and proper log within AI so think long and hard on “securing it without destroying it” with the proper guard rails!