DEV Community

Jane Carter
Jane Carter

Posted on

Cleaning invisible Unicode artifacts from AI-generated text

When text moves between AI chats, editors, and publishing systems, it can carry invisible Unicode characters or typographic artifacts that are hard to spot. These may affect matching, copying, search, or downstream processing even when the text looks normal.

I built Claude Watermark Remover as a small, privacy-first browser tool for cleaning those artifacts from Claude-style AI text. It focuses on normalization and cleanup—not on defeating cryptographic watermarks, and it is not an image watermark remover.

The workflow is intentionally simple:

  1. Paste text into the browser tool.
  2. Review the cleaned output.
  3. Copy it back to your editor or pipeline.

The tool is freemium, and processing is designed around keeping pasted text private. It is useful for writers, developers, and anyone moving AI-generated drafts through systems where invisible characters can cause confusing behavior.

If you have encountered copy/paste issues caused by zero-width characters, non-breaking spaces, or unexpected typography, I would be interested in hearing which cases are most disruptive.

Top comments (0)