Meta announced it will provide its open-source Llama AI models to US government agencies and defense contractors, partnering with major tech and defense companies to integrate the technology into government services.
The collaboration includes tech giants Amazon Web Services, Microsoft, IBM, and Oracle, along with defense contractor Lockheed Martin and data analytics firm Palantir. These companies plan to incorporate Llama into their existing government service platforms.
Several partners have already begun implementing Llama in specific projects, Meta said. Oracle is building a system to analyze aircraft maintenance documents, aiming to cut down on repair times. Scale AI wants to use it for military mission planning and to find weaknesses in enemy defenses. Lockheed Martin has already integrated Llama into its "AI Factory" system.
Meta waves the American flag
In announcing the move, Meta struck a notably patriotic tone, crediting its success to "the entrepreneurial spirit and democratic values the United States."
The company sees applications beyond just military use - Deloitte, for example, is creating Llama-based tools to help government agencies handle education, energy, and small business programs.
And it's also bringing the U.S.-China AI race into the conversation. The company argues that open-source AI models like Llama play a key role in maintaining US technological advantages, essentially as a countermeasure to Chinese open-source modes.
Meta has previously raised concerns about Chinese AI models potentially spreading cultural values to other countries, stating that "the world’s next generation of digital infrastructure" should be "rooted in democratic values and safeguards."
Meta has already used the argument in the past that China is exporting its own cultural values to other societies via AI models. However, the "next generation of digital infrastructure" must be based on "democratic values and security guidelines".
Reuters reported last week that China has been testing Meta's open-source models for military applications.
A separate but related controversy is Meta's use of the term "open source." The Open Source Initiative has challenged Meta's "open source" designation, pointing out that while the company freely distributes its models and weights, it keeps its training data private.
