| Aug 20, 2026 | 🏷 Fanar · Cultural Alignment & Benchmarking Our paper The Alignment Veto: How Safety Training Suppresses Cultural Knowledge in LLMs was accepted at EMNLP 2026 — a diagnostic study of how safety-alignment training systematically suppresses culturally grounded knowledge in large language models. Congratulations to Pardis Sadat Zahraei, and to our collaborators Gokhan Tur and Dilek Hakkani-Tür at UIUC. |
| Jul 10, 2026 | 🏷 Fanar · Document Understanding Our paper Reasoning over Non-Parametric Memory in Multilingual LLMs: A Cognitive-Based Analysis of Relational Knowledge was accepted at COLM 2026 — probing relational reasoning in RAG-style settings across factual, counterfactual, and neutralized contexts, over six languages and Bloom-inspired cognitive levels. Congratulations to Zohreh Abbasi Liasi, Batoul Hammoud, and Anas Madkoor. |
| Apr 19, 2026 | 🏷 Fanar · Image Understanding Two papers arising from our work on Fanar were accepted to Findings of ACL 2026: MorphBPE: Morphology-Aware Tokenization for Efficient LLM Training and Almieyar-Oryx-BloomBench: A Bilingual Multimodal Benchmark for Cognitively Informed Evaluation of Vision-Language Models. |
| Mar 29, 2026 | SilkRoadNLP workshop held at EACL 2026 in Rabat, Morocco — our initiative on NLP for the Iranian family of languages and cultures along the historical Silk Road. |
| Mar 27, 2026 | We are honored to receive the Best Resource Paper Award at EACL 2026 in Rabat, Morocco! |
| Mar 24, 2026 | Arabic NLP School held at EACL 2026 in Rabat, Morocco — a full-day school on foundational and advanced topics in Arabic language technologies, with over 120 participants selected from 300+ applications. |
| Jan 03, 2026 | Three papers accepted at EACL 2026 — two in the main conference and one in Findings. Congratulations to all co-authors! |