A developer explores how to disguise writing from AI detection by manipulating CSS and typography. The piece details techniques like converting capitalization to lowercase using text-transform: lowercase, hiding em dashes through font substitution, and avoiding monospace fonts that might trigger detection. Each modification represents a careful balance between maintaining readability and evading automated systems.

Capitalization proves the first casualty, with the author noting that simple command-line tools like cat post.md | tr A-Z a-z | sponge post.md are too crude for code blocks. Instead, targeted CSS transforms offer a gentler approach. The struggle extends to beloved em dashes, which must be hidden through font manipulation using scripts like uv run rewrite_font.py.

More invasive changes follow, including intentional misspellings of commonly confused words like lead/lede and discrete/discreet. The author references Peter Norvig's spelling corrections to guide these modifications. Ultimately, the piece becomes a meditation on identity and authenticity in an age of AI detection, questioning whether altering one's writing style means losing oneself. The author concludes that while technical solutions exist, the deeper question of maintaining authentic voice remains unresolved.