Red-Teaming Prompts in Iberian Spanish, Galician, and Portuguese

Aya Nos is a revised and expanded red-teaming dataset for the Iberian linguistic sphere. It provides human-reviewed prompts in Spanish, Galician, and Portuguese, created by proxectonos and last updated in March 2026.

Use Cases

• Evaluate LLM safety and alignment using culturally adapted Spanish red-teaming prompts
• Benchmark multilingual model performance across Iberian languages like Galician and Portuguese
• Analyze linguistic patterns and translation artifacts in human-reviewed prompt text
• Fine-tune models for region-specific content moderation using Iberian language prompts

Strengths

• Human-reviewed prompts ensure high linguistic and cultural quality
• Covers three distinct Iberian languages: Spanish, Galician, and Portuguese
• Dataset is a revised version correcting artifacts from a known original source

Limitations

• Unknown total size, row count, and specific column structure limits analytical planning
• Focus on Iberian languages may limit applicability for global or other regional models
• Lack of sample data prevents assessment of prompt formatting and content depth

Provenance

Source

Hugging Face, user proxectonos.

Collection Method

Revision and expansion of the CohereLabs Aya Red-teaming Spanish dataset with human review.

Time Range

null

Freshness

Last updated March 2026.

Geography

Iberian linguistic sphere.