AISP practice · Module 3

AI Data Management, Privacy and Model Security — 15 questions. Choose the BEST answer.

You can answer as a guest, but you'll need a free account to see your score and save it. Sign up · Sign in
Question 1 Question 1 · M3
When is a model registry an effective security control?
Question 2 Question 2 · M3
An API customer sends millions of systematically varied queries and stores all outputs. What is the MOST likely goal, and one control?
Question 3 Question 3 · M3
A data scientist downloads a pickle-format model from a public hub to test it. What is the main risk?
Question 4 Question 4 · M3
Which control BEST ensures the training job uses exactly the approved dataset?
Question 5 Question 5 · M3
Which term describes proof that a dataset was not altered while it moved between teams?
Question 6 Question 6 · M3
In differential privacy, what happens when ε (epsilon) is made smaller?
Question 7 Question 7 · M3
A vector index is built from a mix of Internal and Confidential documents. What classification should the index carry?
Question 8 Question 8 · M3
Staff paste customer account numbers into a public chatbot. Which control is designed to stop this at the point of use?
Question 9 Question 9 · M3
An AI provider's terms let it train its own models on customer prompts. Under privacy law, what does this make the provider for that processing?
Question 10 Question 10 · M3
Which measure reduces both training-data extraction and membership inference risk?
Question 11 Question 11 · M3
A popular ML package release is found to have been compromised and it ran in your build pipeline. Beyond updating the package, what is essential?
Question 12 Question 12 · M3
What distinguishes watermarking from fingerprinting of a model?
Question 13 Question 13 · M3
A team replaces customer names with reversible tokens before fine-tuning and calls the data anonymous. Is that correct?
Question 14 Question 14 · M3
A model passes all accuracy tests but misclassifies any invoice containing a rare watermark as 'approved'. What attack does this suggest?
Question 15 Question 15 · M3
A team wants to rely on cryptographic erase for a decommissioned training cluster. What must be true?