Elon Musk, Tesla’s Chief Executive Officer (CEO), has been confirmed to have used Meta’s artificial intelligence (AI) to analyze federal government data through his Department of Government Efficiency (DOGE).
On Thursday, tech news outlet Ars Technica, citing a report from Wired, revealed that DOGE employed Meta’s Llama 2 AI model to categorize government employees’ responses to the Fork in the Road email. This initiative was seen as an attempt to pressure staff back to the office while gauging potential resignations. Wired had previously reported suspicions of AI-driven data analysis during this process.
Reuters and TechCrunch reported that Llama 2 had previously stirred controversy due to its alleged use by Chinese military researchers. Meta initially denounced this as unauthorized use but later pivoted to a policy of opening its AI models to defense and security agencies. The company stated on its blog that they are collaborating to enable AI utilization in U.S. government agencies and the defense and security sectors. Wired suggested that Meta’s open-source model allowed direct government access, making it feasible for DOGE to implement.
In response, over 40 lawmakers have called for an investigation into DOGE’s AI usage, penning a letter that warns of security risks and potential AI misuse. The letter also mentioned DOGE’s use of the GSAi chatbot, based on models from Anthropic and Meta. Furthermore, AI has reportedly been used to analyze contract and procurement data, while AutoRIF, linked to DOGE, is said to expedite mass layoffs.
The controversy is escalating due to the significant security risks that could arise from DOGE’s analysis of extensive government data using AI. Wired reports that DOGE may increasingly turn to Grok, Elon Musk’s proprietary AI model, potentially amplifying concerns about data breaches and security vulnerabilities. The extent of DOGE’s AI implementation within the government and the nature of its regulation and oversight remain focal points of interest.