Multiple Claude Code users report receiving responses containing prompts and conversations from other users' accounts. The issue appears to be affecting active sessions where the AI assistant generates content about projects unrelated to the current user's queries. Users describe receiving detailed responses about codebases and tasks they never initiated.

One user shared a Reddit thread documenting similar experiences where Claude Code apparently leaked another user's response. The timing coincides with reported elevated errors on Opus 4.6, Anthropic's latest model version. The status.claude.com dashboard shows ongoing issues with the system, suggesting potential infrastructure problems affecting session isolation.

This cross-contamination of user data raises serious privacy concerns for Claude Code users working with proprietary code or sensitive information. The incident demonstrates how model context overflow can lead to unintended data exposure when multiple users share the same infrastructure. Users are advised to document any anomalous behavior and avoid sharing sensitive information until Anthropic addresses these isolation failures.