Addressing the Right to Explanation and the Right to Challenge through Hybrid-AI: Symbolic Constraints over Large Language Models via Prompt Engineering