LLM Copyright Guardrails Easily Bypassed with a ZIP File

ML author Andriy Burkov demonstrated that LLMs refuse to copy copyrighted books from websites but will process the same content if uploaded as HTML/PDF files, showing copyright guardrails only check the source format and can be trivially bypassed.

2026-10-23 ~ 2026-10-23 · 2 related posts

• LLM copyright guardrail absurdity: refuses to scrape a book but happily zips it for you — burkov · 2026-10-23
• Burkov Shows LLM Copyright Guardrail Applies Only to Content Form, Not Substance — burkov · 2026-10-23