This is a dedicated watch page for a single video.
Scenario: A security analyst wants to test an AI system for backdoor vulnerabilities. Which technique should they employ to ensure that the AI continues to perform well on regular inputs but fails under specific conditions crafted by the analyst?