From Sidq to Synthetic Speech: A Culturally Grounded Framework for Ethical Generative AI Communication in Arab Media and Public Relations

Abstract

Generative artificial intelligence (GenAI) is transforming how organizations communicate with their publics. In journalism, public relations, and digital communication, AI systems no longer merely support content production; they participate in message creation, audience interaction, and reputation management. Existing AI ethics frameworks emphasize transparency, accountability, fairness, and privacy, yet offer limited guidance for communication practice in culturally diverse settings. This limitation is especially evident in Arab media environments, where linguistic diversity, cultural norms, and religiously informed moral expectations shape perceptions of communicative legitimacy.

This conceptual article develops the SADIQ Framework, a culturally grounded model for ethical AI-mediated communication in Arab media and public relations. Through critical interpretive synthesis and normative comparison, it connects human-machine communication, communication ethics, AI governance, and contemporary Islamic ethical thought. It organizes the discussion around five tensions: speed versus verification, automation versus accountable agency, personalization versus dignity, optimization versus linguistic and representational justice, and engagement versus public benefit.

The framework proposes five interconnected dimensions: Sidq and tabayyun (truthfulness and verification), Amanah (responsible stewardship), Dignity or karamah (privacy and human agency), Inclusive adl (linguistic and representational justice), and Qasd and maslahah (legitimate purpose and public benefit). All five apply throughout the communication lifecycle. Eight propositions, an organizational audit matrix, and an illustrative scenario specify a research and practice agenda. The contribution is an operational interpretation of ethical commitments for communication workflows; its effectiveness and cultural reception remain open to empirical testing.