An organization's AI ethics policy states that all AI systems must be 'transparent.' A business unit argues their proprietary fraud detection algorithm cannot be disclosed to fraud perpetrators without defeating its purpose. How should this tension be resolved?

B is correct because transparency in AI ethics does not require disclosing all technical details that would undermine the system's purpose — it requires providing affected individuals with enough information to understand decisions affecting them, while allowing proportionate exceptions for legitimate security concerns that are themselves subject to governance oversight. A is impractical and counterproductive. C is too broad an exemption. D limits transparency to one legal framework.