Telecom company Orange France to Employ OpenAI's Artificial Intelligence Models for African Languages
Orange, the French telecom giant, is making strides in democratising Artificial Intelligence (AI) for underrepresented regions, particularly Africa. The company's project could serve as a model for how big tech can partner with local actors to make AI work for these regions.
Historically, African languages have been overlooked in mainstream AI development. This fact was highlighted by a Cornell University study and a report by Nature, which pointed out that over 2,000 African languages have been neglected in the development of AI systems due to data scarcity and limited computational resources.
To address this gap, Orange is actively fine-tuning OpenAI's latest open-weight AI language models to support over 2,000 African regional languages. The company operates in 18 African countries and, earlier this year, began using OpenAI's Whisper speech model for African languages.
By using OpenAI's open-weight models, such as gpt-oss-120b and gpt-oss-20b, Orange can fine-tune the tools with its own African language datasets without needing to start from scratch. This enables the company to support speech recognition, translation, and other advanced natural language tasks in regional African languages.
The deployment of these models could help bring millions of Africans online and into the digital economy by facilitating translation, communication, and access to services in native languages. This initiative is seen as a blueprint for how AI can help bridge the digital divide in Africa.
To comply with regional data regulations and ensure sensitive data protection, Orange hosts these AI models locally across its infrastructure. This careful control of deployment environments supports emerging sovereign AI ecosystems in Africa.
Orange plans to provide the fine-tuned African language AI models for free to local governments and public entities to bridge the digital divide and promote digital inclusion in Africa. Local partnerships play a significant role in building inclusive AI systems, as emphasized by Steve Jarrett, Chief AI Officer at Orange.
The customized models are being developed to support a broad impact on services and society. They help digital inclusion by enhancing regional language voice recognition and translation, optimize network operations, support financial inclusion through AI-powered microloan risk reduction, and contribute to sustainability goals with energy-efficient AI reducing emissions.
Steve Jarrett also highlighted the initiative's role in creating a scalable, ethical AI ecosystem that treats African languages as first-class citizens in AI, addressing enterprise sensitive needs and enabling innovative customer care solutions.
Besides advancing language AI, Orange’s AI strategy includes distillation of models for efficiency leading to substantial emission reductions and extensive digital skills training targets in Africa.
In summary, Orange's initiative represents a pioneering and comprehensive effort to democratise AI for African languages by fine-tuning OpenAI’s models, localizing deployment to respect data sovereignty, and promoting wide societal benefits such as improved communication, financial inclusion, and environmental sustainability across Africa.
Technology, encompassing OpenAI's latest open-weight AI language models, is being tailored by Orange to support more than 2,000 African languages. Artificial Intelligence (AI), as fine-tuned by Orange, is being used to help bridge the digital divide in Africa, promoting digital inclusion and democratizing AI for underrepresented regions.
