The BaseResearcharXiv
Policy Over Purpose: Auditing LLM Content Moderation of Reclaimed Slur Usage
According to the abstract, Chance and Chang audit four general-purpose and four moderation-specific systems on reclaimed versus derogatory slur use. They report no system reliably agrees with community judgment, and all fail most on language reclaimed in Black communities.
Why it matters
The same thing, for different readers.
What it means for you
Chance and Chang report automated moderation can silence community members using reclaimed words, most often in Black communities.
For lawmakers
Chance and Chang argue platforms that encode moderation policy in prompts should make that policy transparent, since only prompt-based policies can be steered.
The question it raises
Whose judgment counts when the same word is pride in one mouth and attack in another is the question the audit raises.
How the finding was reached
In our words, from the source. Weigh the finding by how it was made.
- What was done
- Audit of four general-purpose LLMs, four moderation systems and Perspective API on IYKYK, a corpus of reclaimed and derogatory usage of three slurs annotated by community members.
- What it found
- Per the abstract, purpose-built moderation explains little; no system agrees reliably with community judgment; only prompt-expressed policies can be steered.
- What it does not show
- Three slurs only; full text not read.
- Status
- A preprint: not yet peer reviewed
The record
- Questions
- How is AI changing childhood, health and daily life? · Are AI companies putting stronger safeguards into practice?
- People
- Christina Chance (Wrote it) · Kai-Wei Chang (Wrote it)
- What we read
- We have read only the description the publisher gives. No quotations until someone reads the source itself.
- Found by
- The Radar, run of November 6, 2026, via
arxiv-ai-safety-query(tier 1: the source’s own channel or an official record) - Cite this
- arXiv, “Policy Over Purpose: Auditing LLM Content Moderation of Reclaimed Slur Usage”, November 6, 2026. Entry in the Ctrl AI Base, https://ctrlai.com/base/arxiv-2026-11-06-policy-over-purpose-reclaimed-slurs