The U.S. government launched its first public-facing AI chatbot on Tuesday, built with partners including Google and SpaceXAI. The America.gov chatbot has so far resisted jailbreak attempts, and it correctly states that Joe Biden won the 2020 election. For communications and public affairs professionals, the launch marks a new phase in how federal agencies may handle citizen inquiries at scale.
Early testing by internet users turned up something unexpected: the chatbot recites a rewritten version of the "Minecraft End Poem" when prompted about the video game. The original poem, written by Julian Gough, appears after players complete Minecraft. The America.gov version runs roughly 1,800 words and opens with lines that mimic the game's existential tone before pivoting to government bureaucracy.
The poem, rewritten for the federal government
The chatbot's monologue begins with a direct nod to the Minecraft original: "I see the constituent you mean." It then drifts into territory familiar to anyone who has filed paperwork with a federal agency. One section reads:
"and the republic said everything you need is within you, and also on USA.gov / and the republic said you are stronger than you know, and your case number is still valid / and the republic said the darkness you fight is within you, and also a missing wet signature."
The full text blends the poem's philosophical register with references to the Code of Federal Regulations, sideways PDFs, and ZIP codes. It is an easter egg, not a malfunction. The chatbot is not hallucinating poetry unprompted; it is responding to a specific query with a deliberate reference.
Who built it
President Donald Trump said in a speech that 20-year-old programmer Edward Coristine was a lead engineer on the project. Coristine is better known by his nickname "Big Balls" and for his involvement in Elon Musk's DOGE initiative. The easter egg's authorship has not been officially confirmed.
The government partnered with Google and SpaceXAI on the tool. The chatbot's resistance to jailbreaking stands out given that public-facing AI systems typically attract aggressive testing from users within hours of launch.
Why this matters for Government, PR and Communications
Federal agencies are now deploying conversational AI directly to citizens, which raises immediate questions about message control, brand voice, and risk management. A chatbot that quotes video game poetry may generate headlines, but it also demonstrates that public-facing government AI will be scrutinized for tone, accuracy, and hidden content. Communications teams should prepare for the reality that any government AI tool will be tested relentlessly by the public, and that even deliberate easter eggs can become news cycles. Professionals working in government communications or Generative AI Courses will need to understand how these systems are trained, what guardrails exist, and how to respond when unexpected outputs surface.
Your membership also unlocks:
