It is based on the
aubmindlab/bert-base-arabertv02 base model and fine-tuned on a curated dataset of
Arabic news texts, labeled as either:
The model learns stylistic, syntactic, and semantic patterns that differentiate human journalism from automatically generated text.
1from transformers import pipeline
2
3text = """
4أعلنت وزارة الاقتصاد اليوم عن إطلاق خطة جديدة تهدف إلى دعم الشركات
5الصغيرة والمتوسطة وتعزيز فرص العمل خلال السنوات القادمة.
6"""
7
8classifier = pipeline("text-classification", model="salmane11/konan", tokenizer="salmane11/konan", truncation=True, device = 0)
9
10def detect_ai_generated_news(news: str) -> str:
11 label = classifier(news)
12 if label[0]['label']=="machine":
13 return True
14 else:
15 return False
16#detect_ai_generated_news(aljazeera_news['content'][0])
17
1@article{lamsiyah2025m,
2 title={M-DAIGT: A Shared Task on Multi-Domain Detection of AI-Generated Text},
3 author={Lamsiyah, Salima and Ezzini, Saad and El Mahdaouy, Abdelkader and Alami, Hamza and Benlahbib, Abdessamad and El Amrany, Samir and Chafik, Salmane and Hammouchi, Hicham},
4 journal={M-DAIGT-ST 2025},
5 pages={1},
6 year={2025}
7}
8
9
