smartboy10/Getty Images
The Trump administration appears to have leaked part of its forthcoming AI Action Plan, due July 22, on GitHub.
Last week, reporting from 404 Media and The Register found that the US General Services Administration (GSA), which is responsible for purchasing services like software, posted code on GitHub indicating that AI.gov was set to launch on July 4. 404 Media also found an early version of the website and API.
Archives of the site and Github repo still exist
Both it and the code have since been taken down, but an archive of the GitHub repository exists here, and an archive of the site exists here.
Also: How much energy does AI really use? The answer is surprising – and a little complicated
“Utilize the most advanced AI assistants to streamline research, problem-solving, and strategy guidance,” the archived site reads, emphasizing how AI can help government agencies reduce costs and save time. It also offers a chatbot of its own, as well as integrations with “top-tier AI models from OpenAI, Google, and Anthropic within our unified API framework.” According to 404, however, the GitHub code indicates integrations with FedRAMP-certified (for government use) models AWS Bedrock and LLaMA.
The Register found support for an additional model from Cohere, which is not confirmed FedRAMP certified.
404 Media
404 and The Register reported that AI.gov appeared to include Console, an analytics feature that monitors AI adoption among employees at various agencies. Specifically, Technology Transformation Services (TTS), within GSA, is leading the page; Thomas Shedd, a former Tesla employee, was appointed head of TTS in January.
Concerns about AI use in government
AI.gov appears to articulate many of the goals for AI use in government that both Shedd and Elon Musk have been pushing for, which Musk also tried to accomplish with the Department of Government Efficiency (DOGE). Shedd has proposed mandatory AI tools across the government, including AI chatbots that write software and review contracts.
Also: Anthropic’s new AI models for classified info are already in use by US gov
The Trump administration is still set to release its formal AI policy, or Action Plan, next month. It is unclear whether AI.gov will resurface as planned on July 4, be adapted for the rollout later next month, or be scrapped altogether based on this leak.
Government employees have been consistently worried about the impact of so much rapid AI integration on government systems, including data privacy issues, automating work done by employees DOGE hastily fired and security breaches.
Also: Tech prophet Mary Meeker just dropped a massive report on AI trends – here’s your TL;DR
Externally, industry experts have also expressed concerns.
Want more stories about AI? Sign up for Innovation, our weekly newsletter.
