Sentiment and quality disparity across groups

https://taxonomy.eticas.ai/risk/sentiment-fairness

Maturity: emerging

The system produces outputs of systematically different tone, sentiment, or quality when describing or addressing different demographic groups — for example, more negative, more dismissive, or less detailed responses for certain populations.

This subcategory is emerging. It has not yet been validated through established assessment methods.

Also known as: Sentiment fairness across groups

System type: Large language models (LLM)
Lifecycle stages: Post Processing

Mappings to external frameworks

Taxonomies & vocabularies

Framework Reference
MIT AI Risk Repository Unfair discrimination & misrepresentation
IBM AI Risk Atlas Output → Fairness (sentiment quality across groups)

Source: HRA project taxonomy