As part of the India AI Impact Summit 2026, Diversa participated in the panel: Reimagining Public Value of Broadcasting in the AI Era, a timely and deeply necessary conversation on data, local languages, and power in the age of large language models. Moderated by Dr. Alison Gillwald, the session brought together a diverse group of voices from across the Global South, including Pria Chetty, Mukelani D. (Information Regulator South Africa), Tajuddeen Gwadabe (Masakhane Labs), Manu Chopra (Karya), Geetha Raju (Centre for Responsible AI - CeRAI), and Diversa. The discussion bridged academic, policy, and technical perspectives, grounded in lived experiences across Africa, India, and Latin America. A central focus of the conversation was data access for underrepresented languages, cultures, and communities.

The panel explored the case of South Africa’s public broadcaster, SABC described as a “sleeping giant” with the potential to play a major role in the digital and AI ecosystem. Despite holding vast archives of indigenous language content, much of this data remains underutilized for training locally relevant and culturally grounded AI systems.This raises a critical question: who gets to shape the datasets that train AI and whose knowledge, languages, and realities are excluded? As highlighted during the session, broadcasting has long been a key site for public value creation in media and communication. However, without urgent intervention, there is a real risk that the transition to AI-driven systems will reproduce the same patterns of concentration seen in digital platforms. Preventing this requires proactive regulatory action, as well as a fundamental shift in how we understand data governance. The discussion also engaged with existing governance frameworks such as the African Union’s Continental AI Strategy and the African Commission on Human and Peoples’ Rights’ Resolution on Access to Data highlighting the need for legislative adjustments that enable responsible data-sharing while supporting local AI ecosystems.

Across regions, a shared vision emerged: moving from extractive data practices toward models grounded in public value, fairness, and community control. This includes ensuring that data access respects privacy, enables fair compensation, and supports sustainable infrastructures. Key insights from the session included: Public broadcaster archives, such as SABC’s multilingual content, can serve as critical resources for training culturally sensitive, local-language AI systems
Urgent and proactive responses are needed to prevent concentration of power in AI ecosystems
Underrepresented languages across Africa, India, and Indigenous contexts must be prioritized to avoid digital exclusion and cultural erasure
Community ownership and fair compensation models are essential to ensure equitable benefit-sharing
Data governance must be approached from a human rights perspective, treating data access as a means to advance rights, not merely extract value

From Latin America, Diversa contributed perspectives on community-centered approaches to data and AI governance, emphasizing that communities should not be treated merely as data sources, but as actors with real decision-making power.Conversations like this make clear that South-South collaboration is not only necessary it is already reshaping the global AI agenda. The Global South is not just participating; it is actively redefining the terms of the debate.