The Trump administration’s plans to integrate artificial intelligence into government operations were previewed in a leaked test website on GitHub, ahead of the official AI Action Plan deadline of July 22. The US General Services Administration (GSA) posted code indicating a potential launch of AI.gov on July 4.
The code and early version of the website, since taken down, were initially discovered by 404 Media and The Register. Archives of both the GitHub repository and the website remain accessible. The archived site detailed the use of AI to streamline research, problem-solving, and strategy guidance, emphasizing potential cost and time savings for government agencies.
The archived site featured a chatbot and proposed integrations with AI models from OpenAI, Google, and Anthropic. However, the GitHub code suggested integrations with AWS Bedrock and LLaMA, both FedRAMP-certified for government use, according to 404 Media. The Register identified support for an additional, non-FedRAMP-certified model from Cohere.
404 Media and The Register reported that AI.gov included Console, an analytics feature designed to monitor AI adoption among government employees. Technology Transformation Services (TTS), within the GSA, is leading the initiative, headed by Thomas Shedd, a former Tesla employee appointed in January.
AI.gov reflects goals for AI use in government promoted by Shedd and Elon Musk, including mandatory AI tools like chatbots for software development and contract review. These proposals are similar to Musk’s objectives with the Department of Government Efficiency (DOGE). The potential impact of widespread AI integration on government systems has raised concerns among employees regarding data privacy, automation of tasks, and security breaches.
The Trump administration’s formal AI Action Plan is still scheduled for release next month. The status of AI.gov is uncertain, whether it will be launched as planned on July 4, adapted for the broader rollout, or discontinued.