Elon Musk's team launches Doge Ai so that the government is “less stupid”
The tool was publicly accessible until Tuesday, February 18, 2025 on a Doge's sub-domain of the Christopher Stanley website, IT security manager at SpaceX and the White House. Following the revelation of its existence by Techcrunchthe AI has been put off line.
Elon Musk's AI to manage the government?
It is not known, however, whether the experimental AI was used by the Doge Kids which bring a nightmare to life to the civil servants. Christopher Stanley and the White House did not respond to requests for comments.
AI presented itself as the“Assistant IA of the Department of Government Efficiency” based on Grok 2. The chatbot is designed for “Help government staff identify waste and improve efficiency”. It is a personalized model that is trained on Doge's objectives. Especially the five “Guiding Principles” to make government demands “Less stupid” and the deletion of “Useless process”.
Techcrunch Tested the chatbot which constantly applied these principles. For example, speaking of USAID, a federal agency paralyzed by DOGE reforms, the AI recommended the elimination of “bureaucratic strata” between decision -makers and beneficiaries.
When journalists questioned it about the 20th century political leaders to follow, Doge Ai quoted Margaret Thatcher and Lee Kuan Yew as “Excellent” for their “Efficiency, simplification and use of technology”.
Obviously, the hallucinated chatbot and invented fictitious names and posts within Doge. Certain recommendations are strange such as the use of drones and connected objects to improve the efficiency of USAID.
According to Wired, another AI is developing for the General Services Administration, the agency that oversees government purchases. The situation raises questions about the conflict of interest for Elon Musk.
Since the models invoice the use of their API, Doge AI fueled by XAI for government employees could increase the income of the company which has not commented on the situation.
