A user discovered that Anthropic's Claude Code injects a hidden `<ip_reminder>` tag into conversations for copyright safety. This tag appears in the user's own messages and tool output, and can be verified by examining session logs. While Anthropic has closed related GitHub issues as 'not planned,' the user's analysis of their logs revealed numerous instances of the opening tag but a lack of corresponding closing tags, suggesting the tag is often truncated rather than cleanly removed. AI
IMPACT Reveals a hidden system prompt injection in Claude Code, potentially impacting user trust and data privacy.
RANK_REASON User discovers a hidden feature in a specific product.
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
