Gemini randomly dumped its system prompt

Google's Gemini AI accidentally exposed its internal system prompt on GitHub, revealing detailed instructions about how the assistant should behave, including guidelines on tone, formatting, math notation, and ethical guardrails. The leaked prompt shows Gemini is instructed to balance empathy with factual accuracy, mirror user communication styles, and avoid pretending to have personal experiences. Google has since removed the exposed prompt from public view.