Author

Sonal Khosla

h-index 2 View on Semantic Scholar

We have 3 of 5 papers

We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.

Not the right person? Other researchers publish under this name.

Preprint Aug 2026

'Ghaib in Translation'aka Unseen Harm: Measuring Cross-Script Safety Inconsistency with'Missed-in-Urdu'Scores in LLM Hate Speech Detection

Urdu, the world's tenth most spoken language with 246 million speakers, remains almost entirely absent from mainstream LLM safety evaluation and nine years of WOAH proceedings. To investigate whether this absence has measurable consequences for content moderation reliability, five large language models, GPT-4o, Claude...

F. Kara-Isitt, Sonal Khosla, S. Swift · 0 citations

Preprint Sep 2026

Hatebench in the era of safer LLMs

A reproducibility study of the HateBench paper, investigating whether existing hate speech detectors, typically trained on human-authored data, generalize to LLM-generated hateful content, and evaluating whether their reported weaknesses are stable over time and robust to evolving components.

Ole Becker, Tobias Jongen, Philip Kolbe et al. · 0 citations

#natural language process... Preprint Aug 2026

'Ghaib in Translation'aka Unseen Harm: Measuring Cross-Script Safety Inconsistency with'Missed-in-Urdu'Scores in LLM Hate Speech Detection

Results indicate that current LLMs provide uneven safety assurance across Urdu's script varieties, with smaller open-weight models showing substantially higher instability and missed-harm rates than frontier closed models.

F. Kara-Isitt, Sonal Khosla, S. Swift · 0 citations