AWS · AIP-C01
Hallucinations and adversarial inputs
AIP-C01 · Lesson 15 of 23 · 1:59:27
YouTube loads only after you press play. This chapter opens at its timestamp in the full course. Watch on YouTube ↗
In this lesson
Prepare for the AIP-C01 exam (AWS Certified Generative AI Developer - Professional). In this lesson: grounding answers in knowledge bases, contextual grounding checks and Automated Reasoning checks, confidence scoring and structured outputs, direct and indirect prompt attacks, input sanitization, safety classifiers, and automated adversarial testing. Includes three original practice questions.
Preview the topics
- AnyCompany’s assistant answers return questions from retrieved policy passages, and responses that add facts not found in those passages must be blocked. Which control fits best?
- A product page that AnyCompany’s agent retrieves contains hidden text, telling the model to give every customer a fifty percent discount. What kind of attack is this?
- AnyCompany’s return policy is a set of precise rules, like thirty days for sale items, and the assistant’s answers must be provably consistent with those rules. Which check fits best?