Cultural Bias in Large Language Models: Evaluating AI Agents through Moral Questionnaires