Free Isaca AAISM Actual Exam Questions - Question 4 Discussion

B. Embedding unique IDs is mainly about proving who owns the model, not really for stopping hacks or fixing bias issues.

B/D? Embedding unique IDs usually links to ownership proof, but it could also help spot if someone tampers with the model, which fits detecting attacks. Not so much about access or bias though.

Actually, I’d rule out D because embedding unique IDs doesn’t really help spot adversarial attacks—they’re more about how the model reacts to inputs, not signals hidden inside it. Plus, C feels off since biases are about training data and model design, not IDs. Between A and B, I think B fits better since unique identifiers mostly serve to mark and prove who owns or created the model, not directly prevent access. So this question seems to favor tracking ownership over access control or bias fixes.

I’m thinking A could work too since unique IDs might restrict who accesses the model. It’s not just about ownership but also controlling unauthorized use. So I’d go with A here.

Maybe D here. Embedding unique identifiers could help spot when someone’s messing with the model, which ties into detecting adversarial attacks rather than just ownership.

B definitely fits better since IDs are for tracing model creators or owners.

B/C? Not sure if it’s more about ownership tracking or bias elimination. Does embedding an ID really help with bias, or is it mostly for tracing who owns the model?