Where is AI-based threats in the cyber landscape

One thing I'm a bit confused on is AI security's location in the security umbrella. When we talk about protecting against AI threats, are we talking about application security and developing secure code, are we talking about protecting the data it's trained on and what it gives out, which would fall under more data security? Or are we talking more about protecting the infrastructure and networking that the systems running the models use, which is more network/cyber security? Or a …

ai security ai threats application application security code cyber cyber landscape cybersecurity data location protecting secure code security talking threats umbrella under