AI red teaming (prompting) and flaws evaluation (student skills)
You participated in AI red teaming activities, focusing on identifying weaknesses and testing model behavior through adversarial probing. You also worked on flaws-related evaluation to understand how the AI system fails under specific conditions. The activities were aimed at improving prompt safety and model robustness through feedback-style analysis. • Performed adversarial tests against AI responses using red-team prompting • Identified and documented failure modes and vulnerabilities • Evaluated outputs to detect common flaw patterns • Used results to inform improvements to prompts and model behavior