French telecom provider Orange has partnered with OpenAI, becoming the first European telecom operator to access pre-release versions of OpenAI’s AI models. This multi-year deal positions Orange as a leader in AI integration, providing it with the unique opportunity to influence OpenAI’s future developments.
As part of this collaboration, Orange will also work with Meta to expand AI-driven translation services in West Africa.
Starting in 2025, the company will leverage regional languages Wolof and Pulaar, spoken by millions, to train Meta’s Llama and OpenAI’s Whisper models for customer service. These open-source models will be accessible to governments, educational institutions, and startups for non-profit applications in health and education, aiming to bridge the digital divide.
In the press release, Orange emphasized that the data management and hosting for its AI models will be conducted in secure European data centers. This aligns with the growing trend of “sovereign AI,” ensuring data storage and processing are localized.
The approach respects regional languages, cultures, and histories while ensuring compliance with European data sovereignty regulations.
Steve Jarrett, Orange’s Chief AI Officer, highlighted the company’s commitment to responsible AI use. He emphasized that while large language models offer significant benefits, they will be deployed strategically to manage energy consumption.
Additionally, Orange’s AI tools will support over 50,000 employees and provide AI-driven voice assistance to enhance customer service.
This partnership aligns with global trends, where regions seek to control their AI infrastructure. By localizing data and focusing on regional solutions, Orange aims to lead in both technological innovation and ethical AI practices.
