• Meta rolls out Llama AI to US defence agencies and allies.
• Open-source design lets governments use it securely with their own data.

Meta’s AI work with the US military is moving beyond experiments and into deployment, with its Llama models now being tested for defence and national security purposes.

Last year, the company first introduced Llama 3.2 with its smaller on-device models and multi-modal systems, saying it was designed to be useful to enterprises and public agencies. Executives explained at the time that government adoption is a natural step, given how researchers, developers, and companies are already experimenting with Llama in different sectors.

Where Meta has placed particular emphasis is on how Llama fits the security needs of government users. The company has argued that its open-source models are especially well-suited for defence applications because they can be downloaded, hosted, and fine-tuned in secure environments.

Unlike many commercial AI services, Llama does not require sensitive data to pass through third-party providers. This makes it possible for agencies to train the models with classified information, run them on secure systems, and deploy them directly on field devices when needed.

Since late last year, the models have been available not only to US government agencies but also to America’s “Five Eyes” partners – Australia, Canada, New Zealand, and the UK – along with private contractors supporting their work. Access is now being widened to key allies in Europe and Asia, including France, Germany, Italy, Japan, and South Korea, as well as to NATO and European Union institutions. Meta says this measured rollout reflects the importance of ensuring that allies share access to advanced AI while maintaining security controls.

To deliver these solutions, Meta is working with a wide range of companies in the defence and technology sectors. Its list of partners includes Accenture, AWS, AMD, Anduril, Ask Sage, Booz Allen, C3 AI, Databricks, IBM, Microsoft, Lockheed Martin, Oracle, Palantir, Scale AI, and Snowflake. Partners help adapt and integrate Llama models into secure environments and specialised systems designed for government use.

Some of that work is already underway. In partnership with the US Army’s Combined Arms Support Command, Meta has been running a pilot to demonstrate how AI and extended reality technologies could improve equipment maintenance. The project explores how tools like Llama, combined with VR and AR, might speed up routine repairs and return vehicles and machinery to service more quickly – potentially reducing downtime in the field.

Meta is also collaborating with Anduril on wearable technologies, including VR and XR headsets, to give soldiers enhanced perception and decision-making support. The company describes this as one of the largest efforts to equip US troops with advanced hardware designed to complement AI-powered systems.

The framing of the initiatives goes beyond immediate military applications. Meta has linked its work to broader geopolitical and economic realities, where technological leadership is closely tied to national security and international influence. The company has pointed to the US government’s AI Action Plan for America as a signal that open-source models like Llama will play a central role in maintaining allied leadership in AI development.

Meta has also stressed the importance of ethics and law in this expansion. It has been said that countries using AI for defence must ensure it is deployed responsibly and in line with international standards. The United States and many of its allies have already committed to these principles through the Political Declaration on Responsible Military Use of Artificial Intelligence and Autonomy, which sets guidelines for how AI should and should not be applied in conflict.

Meta says it is taking a step-by-step approach to extending access to Llama for defence and national security. Each expansion is undertaken in consultation with the US government, and the company has left open the possibility of adding further countries in the future.

The steady growth of Llama in this space signals that open-source AI is becoming an embedded part of defence planning and strategy among democratic allies.

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and co-located with other leading technology events. Click here for more information.

AI News is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.