Details are dribbling out around the Trump administration’s imminent launch of an initiative to push AI across the federal government.
The sweeping AI.gov website, as the initiative is dubbed, briefly popped up on a GitHub repository before it was pulled down. From what was briefly seen, this much has been gleaned: The General Services Administration (GSA) and its Technology Transformation Services (TTS) group would oversee the effort of government agencies adding artificial intelligence (AI) to their day-to-day operations when the initiative launches on July 4.
TTS chief Thomas Shedd, a former software integration engineering manager at Tesla Inc. and ally of Elon Musk, reportedly wants GSA to operate like a software startup with an AI-first strategy to automate much of the work done by federal employees. This, in turn, has led to widespread concern about job security and data handling among industry experts, privacy advocates, and thousands of federal workers.
For now, the AI.gov project has three loosely defined goals: a chatbot; an “all-in-one API” that would allow agencies to connect their systems to models from OpenAI, Anthropic, and Alphabet Inc.’s Google; and “CONSOLE,” nebulously described on the page as a “groundbreaking tool to analyze agency-wide implementation.” Translation: CONSOLE will let agencies monitor AI use in real time to observe which tools federal employees are using.
AI.gov will serve AI models via Amazon Bedrock and, perhaps, enterprise AI firm Cohere, according to API documentation from the GitHub page. The page also suggested AI.gov will rank models.
“What we are seeing now is something that has been occurring for a long time, but just more acute now,” tech expert and privacy advocate Katalin Bártfai-Walcott said in an interview. “These guys are collecting data at an accelerated rate. It is very, very hard to undue this.”
But with the Trump administration, anything can happen–and has–especially with the tech industry. Under AI czar David Sacks and various federal agencies, the administration has veered from punitive regulation (Google, Meta Platforms Inc.) and outright threats (Apple Inc. via tariffs and demands to build iPhones in the U.S.) to beneficiary (the $500 billion Stargate project for OpenAI, Oracle Corp.) and protector (the elimination of state-level AI regulation for a decade under the One Big Beautiful Bill Act).
The case study of the dysfunction is the relationship between Trump and Musk. In the latest wrinkle to their love-hate spat, Musk took to social-media platform X on Wednesday to apology as Trump threatens to withdraw billions of dollars in government contracts to SpaceX as well as punish Tesla with reduced EV credits.
What the government does with AI as an automation tool is particularly worrisome to security experts, civil libertarians, and federal workers, who fear they will be replaced and/or diminished by the technology. The Department of Government Efficiency (DOGE) has also shown a willingness to peruse personal information indiscriminately. Earlier this week, the U.S. Supreme Court gave DOGE access to Social Security systems containing personal data on millions of Americans.
