The GSMA and Zindi on Wednesday unveiled the African Trust & Safety LLM Challenge, a landmark initiative to define next-generation global standards for trustworthy artificial intelligence, with a focus on Africa’s complex linguistic and cultural environment.
The announcement came at the Mobile World Congress 2026 in Barcelona, positioning Africa as a critical testing ground for AI safety amid the rapid adoption of generative AI across sectors such as healthcare, financial services, telecommunications, education, and government platforms.
“With more than 2,000 languages, widespread multilingualism, and culturally nuanced communication, Africa presents a uniquely rigorous environment for AI,” said Celina Lee, CEO and co-founder of Zindi. “Through this challenge, we are placing African AI talent at the forefront of shaping global standards for safe and reliable AI.”
The competition, running from March 4 to April 19 on the Zindi platform, leverages a network of over 100,000 data scientists and AI practitioners from 180 countries. Participants will develop structured adversarial prompts and safety classifications to stress-test large language models (LLMs) trained or deployed in Africa, especially in underrepresented languages and code-switched contexts.
Outputs from the challenge will contribute to an Africa-focused AI trust and safety benchmark, creating evaluation tools designed for real-world, multilingual, and culturally diverse applications. Organisers said the initiative aims to reduce AI-related harms, strengthen digital trust, and inform global AI governance frameworks.
More Read
Louis Powell, Director of AI Initiatives at GSMA, emphasized the stakes for Africa’s mobile ecosystem. “Safety and reliability are paramount as AI adoption accelerates. This collaboration supports the creation of practical benchmarks that reflect Africa’s linguistic diversity and deployment realities, unlocking the potential for inclusive digital growth.”
The challenge offers a total prize pool of $5,000 and is open to participants across Africa and globally. Organisers noted that while most AI safety frameworks are built around a limited set of dominant languages, the African context—with its rich linguistic and cultural diversity—offers a critical proving ground for robust AI systems that can perform safely in emerging markets worldwide.
Analysts said the initiative could also have broader implications for global AI development. By systematically identifying vulnerabilities in African-trained or Africa-deployed models, the challenge could inform safer deployment practices, better model evaluation standards, and more culturally aware AI governance beyond the continent.
More Read
Africa’s growing digital economy, driven by expanding mobile connectivity and increasing adoption of AI technologies, has highlighted the need for regionally relevant AI evaluation. According to Zindi and GSMA, this initiative ensures African expertise directly contributes to global conversations on AI trust, safety, and accountability.
The African Trust & Safety LLM Challenge represents a first-of-its-kind approach to AI evaluation, combining open competition, practical model testing, and policy-oriented outputs. Organisers hope it will foster innovation, empower local talent, and ensure AI technologies deployed across Africa are reliable, fair, and safe for diverse communities.
The challenge underlines Africa’s emerging role in the global AI landscape, not only as a consumer of technology but as a leader in shaping safe and culturally competent AI standards.
