Recent reports indicate that the Department of General Services (DOGE), operating under the Office of Personnel Management (OPM), is leveraging AI technology to enhance its operational efficiency. Specifically, DOGE utilized Meta’s Llama 2 model for categorizing responses from federal employees pertaining to a communication titled “Fork in the Road.” Interestingly, despite the availability of the more advanced Llama 3 and Llama 4 models, DOGE opted for the earlier version. At the time of this initiative, Elon Musk’s Grok model remained proprietary and inaccessible for this function.
Moving forward, there is speculation that DOGE may increasingly integrate Grok into its processes. Notably, Microsoft recently announced its plans to host xAI’s Grok 3 model on the Azure AI Foundry platform, thereby broadening access to advanced AI systems for various applications. However, this pivot raises critical concerns. Over 40 legislators have voiced alarm regarding the transparency and security implications of deploying AI to handle sensitive governmental emails, urging for further scrutiny from the Office of Management and Budget. The potential risks include unauthorized data access and challenges in the legitimate deployment of AI within federal frameworks.