Bypass copy protection in PDFs

12 Mar, 2026 • 276 words • 2 min

Problem statement: Open contract.pdf in Preview.app in macOS. Copy a text portion in it. A dialog pops up:

Without the owner password, you do not have permission to copy portions of the document.

Enter the password to unlock copying from the document.

Excuse me? I do not need your permission.

Attempt #1: Claude #

Open Claude Code. Prompt:

output it in markdown, translated to English

output a summary in the beginning, then output the full translation afterwards

It gets stuck. Boooo!

There’s nothing more annoying than a PC¹ LLM that fails silently without telling you why.

Attempt #2: Codex #

Open Codex. Use the same prompt.

• I'm locating Vertrag-Entwurf.pdf and extracting its text first so I can translate it accurately and format the result in
  Markdown.

• Explored
  └ List rg --files -g Vertrag-Entwurf.pdf
    List Vertrag-Entwurf.pdf

• The file is present. I'm extracting the PDF text now and checking whether the layout needs OCR-style cleanup before translation.

• Ran which pdftotext
  └ /opt/homebrew/bin/pdftotext

• Ran pdfinfo /Users/thiago.perrotta/Downloads/Vertrag-Entwurf.pdf
  └ Creator:         Chromium
    Producer:        {redacted}
    … +14 lines
    Optimized:       no
    PDF version:     1.4

• The PDF has text content but copy protection is enabled, so I'm testing extraction directly. If that fails, I'll fall back to OCR
  from rendered pages.

• Ran pdftotext /Users/thiago.perrotta/Downloads/Vertrag-Entwurf.pdf -
  └ {redacted}

    … +1264 lines

• # Summary

[...]

# Full Translation

[...]

This is why we need competition.

This is why we need diversity.

In this context, OpenAI is the non-PC model.

Do you see the problem with arbitrarily defining what is and isn’t PC? Who gets to have this privilege? Why? Is it universal?

DRM in PDFs suck.

Politically correct. ↩︎

¬ just serendipity 🍀 (not just serendipity)