A transparency discovery that sparked discussion across the tech community

Claude Code is Steganographically Marking Requests

Based on Hacker News Story

2025

1,925

Hacker News Points

A viral story revealing hidden markers in Claude Code prompts

03

The Discovery

Uncovering hidden markers in AI tool interactions

What Was Found

01

Steganographic Markers Claude Code embeds hidden text markers in user requests
02

Invisible to Users These markers are not visible during normal usage
03

Tool Identification Markers identify requests originating from Claude Code
04

Source Attribution Enables tracking of tool-specific interactions

How the Marking Works

Steganographic Marker Structure

// Example of hidden marker structure
// (Conceptual representation)
{
  "marker": "CLAUDE_CODE_TAG",
  "version": "1.0",
  "timestamp": "...",
  "session_id": "..."
}

Hidden markers are embedded in requests to identify Claude Code usage

06

Why This Matters

Privacy, transparency, and trust in AI tools

Key Concerns

visibility_off

Transparency

Users unaware of tracking mechanisms

security

Privacy

Questions about data collection practices

verified_user

Trust

Need for clearer disclosure in AI tools

gavel

Ethics

Balancing utility with user consent

Thank You

The conversation about AI transparency continues

Source: thereallo.dev/blog/claude-code-prompt-steganography