Meta has announced the expansion of access to its Llama AI models for key US allies, supporting defence and national security initiatives.
Llama, an open-source platform, allows governments to securely deploy and fine-tune models using sensitive data without relying on third-party AI providers.
“Llama is particularly well-suited to these sensitive use cases because, as an open source platform, it can be securely downloaded and deployed without the need to transfer sensitive data through third-party AI providers,”
Meta said.
Since last year, Llama has been available to US government agencies and the Five Eyes partners, Australia, Canada, New Zealand, and the UK, and their private sector collaborators.
The expansion now includes France, Germany, Italy, Japan, and South Korea, as well as NATO and European Union institutions.
Meta noted that Llama has been used to enhance decision-making, mission-specific capabilities, and operational efficiency in US military and national security contexts.
For instance, the Army’s Combined Arms Support Command is piloting the use of AI with augmented and virtual reality technologies to accelerate routine equipment repairs.
To support allied deployment, Meta is working with partners including Accenture, Amazon Web Services, AMD, Anduril, Booz Allen, C3 AI, Google Cloud, IBM, Microsoft, Lockheed Martin, Oracle, Palantir, and others.
The company is also developing wearable augmented and virtual reality technologies with Anduril to improve perception and decision-making for US soldiers.
Meta emphasised the importance of responsible AI use in defence.
“It is the responsibility of countries leveraging AI for national security to deploy AI ethically, responsibly, and in accordance with relevant international law and fundamental principles,”
the company said, referencing the Political Declaration on Responsible Military Use of Artificial Intelligence and Autonomy.
Meta plans a step-by-step approach to extending Llama access and will consider further countries in consultation with the US government.
Featured image credit: Meta
