![[2602.13504] From Perceptions To Evidence: Detecting AI-Generated Content In Turkish News Media With A Fine-Tuned Bert Classifier](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.13504] From Perceptions To Evidence: Detecting AI-Generated Content In Turkish News Media With A Fine-Tuned Bert Classifier
Summary
This study presents a fine-tuned BERT classifier for detecting AI-generated content in Turkish news media, achieving a high F1 score and revealing significant AI usage.
Why It Matters
As AI-generated content becomes more prevalent, understanding its impact on media integrity is crucial. This research fills a gap in empirical studies on AI content in Turkish journalism, providing data-driven insights that can inform future media practices and policies.
Key Takeaways
- The study fine-tunes a BERT model specifically for Turkish news media.
- It achieves a 0.9708 F1 score, indicating high accuracy in detecting AI-generated content.
- Approximately 2.5% of analyzed articles were found to be AI-rewritten.
- This research is the first empirical investigation into AI content in Turkish journalism.
- The findings highlight the need for ongoing monitoring of AI's role in news media.
Computer Science > Computation and Language arXiv:2602.13504 (cs) [Submitted on 13 Feb 2026] Title:From Perceptions To Evidence: Detecting AI-Generated Content In Turkish News Media With A Fine-Tuned Bert Classifier Authors:Ozancan Ozdemir View a PDF of the paper titled From Perceptions To Evidence: Detecting AI-Generated Content In Turkish News Media With A Fine-Tuned Bert Classifier, by Ozancan Ozdemir View PDF HTML (experimental) Abstract:The rapid integration of large language models into newsroom workflows has raised urgent questions about the prevalence of AI-generated content in online media. While computational studies have begun to quantify this phenomenon in English-language outlets, no empirical investigation exists for Turkish news media, where existing research remains limited to qualitative interviews with journalists or fake news detection. This study addresses that gap by fine-tuning a Turkish-specific BERT model (dbmdz/bert-base-turkish-cased) on a labeled dataset of 3,600 articles from three major Turkish outlets with distinct editorial orientations for binary classification of AI-rewritten content. The model achieves 0.9708 F1 score on the held-out test set with symmetric precision and recall across both classes. Subsequent deployment on over 3,500 unseen articles spanning between 2023 and 2026 reveals consistent cross-source and temporally stable classification patterns, with mean prediction confidence exceeding 0.96 and an estimated 2.5 percentage of e...