# Training large language models on narrow tasks can lead to broad misalignment

> Research article (Nature, 2026) · cited 10× · AI/ML

**Wikidata**: [openalex:W7124177090](https://www.wikidata.org/wiki/openalex:W7124177090)  
**Source**: https://4ort.xyz/entity/training-large-language-models-on-narrow-tasks-can-lead-to-broad-misalignment