The risks of non-secure AI are growing. Security for AI needs to include concepts from traditional IT security but the nature of new machine learning methods and architectures brings new risks and therefore a need for new solutions.
Francis Rose of Govmatters.TV interviewed Bob Gourley of OODA LLC on what is similar and different about risk mitigation for AI.
Risks from AI include ethical concerns, like unfair and biased systems, but also attacks that can be launched by adversaries to change algorithms or corrupt data.
Extensive research is being applied to mitigating these challenges.
The interview is available at this link and embedded below:
The megatrend of AI and its business impact is examined in more detail in the CTOvision Report on Artificial Intelligence.