Nick Clegg, president of global affairs at Meta, confirmed the move in a blog post, suggesting that open source large language models like its Llama series “can support many aspects of America’s safety and national security.”
“[Llama models] can help to streamline complicated logistics and planning, track terrorist financing or strengthen our cyber defences,” the former British Deputy Prime Minister wrote.
Subscribe today for free
Meta is bringing its Llama models — including the recently released Llama 3.2 — to US government agencies through its army of external partners, including AWS, IBM, Palantir, Microsoft, and Accenture Federal Services, among others.
Among the early security-related applications Llama is being used for include Oracle users applying the models to synthesise aircraft maintenance documents to help technicians speed up repair times, and Scale AI is fine-tuning one of the Llama models to support national security team missions, such as planning operations and identifying adversaries’ vulnerabilities.
Lockheed Martin, meanwhile, is using the Llama models as part its AI Factory, a project aimed at accelerating the development and integration of AI across its aerospace and defence products.
In Meta's blog post, Clegg suggested that use of the Llama models for security applications will "not only support the prosperity and security of the US, they will also help establish US open source standards in the global race for AI leadership.”
“In a world where national security is inextricably linked with economic output, innovation and job growth, widespread adoption of American open source AI models serves both economic and security interests,” Clegg wrote. “Other nations, including China and other competitors of the US, understand this as well, and are racing to develop their own open source models, investing heavily to leap ahead of the US.”
The news came mere days after a Reuters report claiming Chinese military researchers have repurposed an earlier version of Llama to power AI military tools such as to gather and process intelligence.
“We believe it is in both America and the wider democratic world’s interest for American open source models to excel and succeed over models from China and elsewhere,” Clegg wrote.
Meta’s description of its Llama herd of AI models as open source goes against the recently published definition of what constitutes an open source AI system.
The definition, created by the nonprofit Open Source Initiative (OSI) group, states that AI systems require information on their underlying training data to be published to be considered open source — a practice Meta does not employ for its Llama models.
RELATED STORIES
Meta unveils Llama 3.2: Smaller AI models for edge and mobile devices
Meta reports massive spike in cloud demand for its Llama AI models
Meta taps Lumen to boost network capacity for AI service expansions