I banned the em dash from my AI's voice

Description and use

The author removed all em dashes from AI-generated content by adding a token-based ban rule in the rewrite tool, a pre-deploy gate script, and updating the AI prompt. This improved style consistency and prevented machine-like prose.

Problem

The AI-generated content repeatedly used em dashes, a stylistic marker that the author never uses and that signals machine prose.

Outcome

The author eliminated all em dashes from AI-generated content, improving stylistic consistency and preventing machine-like prose. The gate prevented any future violations, ensuring compliance automatically.

Workflow

• 1. Added a banned-words rule to hyphos with token mode for em dashes.
• 2. Updated the rule to match literal dash tokens and flag them.
• 3. Implemented a shell script (no-dash-check.sh) to scan all content before upload.
• 4. Added the ban to the AI prompt for the weekly digest.
• 5. Rewrote existing drafts, replacing each dash with appropriate punctuation (colon, comma, semicolon, new sentence).
• 6. Tested the rule with self-tests and site gate to ensure no dashes slip through.

Evidence

The rule matched on word boundaries, and a dash is not a word, so it could never fire.

Source

The gate proved itself within a day, and against me. On the first of September a description line on one page carried a single dash, and every deploy for the next twenty-four hours refused.

Source