Zeabur breach leaks users' AI API keys; founder apologises
Taiwan's Zeabur told users it detected unauthorised access to an internal service credential on 27 August, exposing stored environment variables including OpenAI, Anthropic, OpenRouter and Gemini API keys. Users reported surging Anthropic and xAI charges, and AlphaMatch reported at least one had been told by support there was "no anomaly". The founder apologised and promised compensation; suspicious LiteLLM-related activity is under investigation.
Original report: Zeabur breach leaks users' AI API keys; founder apologises