Meta supports national security with Llama AI for US agencies

Meta is expanding the reach of its AI models, making its Llama AI series available to US government agencies and private sector partners involved in national security projects. Partnering with firms like Lockheed Martin, Oracle, and Scale AI, Meta aims to assist government teams and contractors with applications such as intelligence gathering and computer code generation for defence needs.

Although Meta’s policies generally restrict using Llama for military purposes, the company is making an exception for these government partners. This decision follows concerns over foreign misuse of the technology, particularly after reports revealed that researchers affiliated with China’s military had used an earlier Llama model without authorisation for intelligence-related applications.

The choice to integrate open AI like Llama into defence remains controversial. Critics argue that AI’s data security risks and its tendency to generate incorrect outputs make it unreliable in military contexts. Recent findings from the AI Now Institute caution that AI tools could be misused by adversaries due to data vulnerabilities, potentially putting sensitive information at risk.

Meta maintains that open AI can accelerate research and enhance security, though US military adoption remains limited. While some big tech employees oppose military-linked projects, Meta emphasises its commitment to strengthening national security while safeguarding its technology from unauthorised foreign use.