Comment by cat-whisperer

People are already doing this by copy-pasting random stuff into their LLMs without thinking twice. I think the fixed number vs. percentage thing makes it way more practical for attackers. Would be cool to see defenses at the data ingestion layer!