Mozilla released its State of Open Source AI report in July 2026, showcasing diverse global projects using open source AI models. Examples include a Māori broadcaster in New Zealand training speech models for te reo, PwC fine-tuning finance language models for clients, and East African farmers diagnosing cassava disease with offline phone-based AI. The report emphasizes community ownership and open licensing as key themes.
The report’s CTO, Raffi Krikorian, highlighted projects such as a Lausanne-based team collaborating with the Red Cross on an open medical model aligned with humanitarian guidelines, preparing for clinical trials in Switzerland and Tanzania. In Switzerland, a public consortium trained a national AI model on public supercomputers, releasing all components openly. These initiatives operate independently without renting cloud resources, underscoring ownership and control by the communities involved, according to stateofopensource.ai.
Mozilla draws parallels to its origins defending open web access 25 years ago, framing the current AI landscape as a similar battleground between open communities and centralized control. The report argues that open source AI fosters competition and innovation, enabling tailored solutions for niche languages and underserved regions. This contrasts with proprietary AI models that often impose usage fees and limit data control, highlighting the importance of open licensing and community governance.
The State of Open Source AI report, version 1.0, was published in July 2026 by Mozilla and is available at stateofopensource.ai. It documents multiple projects across continents, illustrating the growing impact of open source AI in sectors from language preservation to agriculture and healthcare.