Publication
Framing Generative AI Governance in Online News
A timeline analysis from ChatGPT's launch to the EU AI Act (2022–2026)
Brewen Couaran, Yuvraj Singh Pathania, Arjun Rajesh Nair
A framing analysis of how online news shaped public understanding of generative-AI governance from ChatGPT's November 2022 launch through June 2026, tracking six governance frames across over 1.1M multilingual articles.
This study examines how online news shaped public understanding of generative-AI governance in the four years following the public launch of ChatGPT. Using the GDELT 2.0 Global Knowledge Graph as the article source, a two-stage method first performs multilingual keyword matching and then validates candidate matches with LaBSE sentence embeddings, yielding a corpus of 1,116,091 articles scored across six governance frames: Innovation & Opportunity, Risk & Safety, Regulation & Governance, Rights & Privacy, Economic Competition & Labour, and Misinformation & Integrity.
The analysis reveals a clear discourse shift — from innovation-oriented framing in 2022–2023 toward risk- and regulation-focused framing in 2024–2026. An event study of major policy milestones finds an asymmetric media response, and regional comparison shows US and EU outlets emphasizing different frames.
The full pipeline, the scored dataset, and an interactive companion site are released alongside the paper — see the companion project GenAI-GDELT for the code, dataset, and deployed site.
A framing analysis of how online news shaped public understanding of generative-AI governance from ChatGPT's November 2022 launch through June 2026, tracking six governance frames across over 1.1M multilingual articles.
This study examines how online news shaped public understanding of generative-AI governance in the four years following the public launch of ChatGPT. Using the GDELT 2.0 Global Knowledge Graph as the article source, a two-stage method first performs multilingual keyword matching and then validates candidate matches with LaBSE sentence embeddings, yielding a corpus of 1,116,091 articles scored across six governance frames: Innovation & Opportunity, Risk & Safety, Regulation & Governance, Rights & Privacy, Economic Competition & Labour, and Misinformation & Integrity.
The analysis reveals a clear discourse shift — from innovation-oriented framing in 2022–2023 toward risk- and regulation-focused framing in 2024–2026. An event study of major policy milestones finds an asymmetric media response, and regional comparison shows US and EU outlets emphasizing different frames.
The full pipeline, the scored dataset, and an interactive companion site are released alongside the paper — see the companion project GenAI-GDELT for the code, dataset, and deployed site.
Companion Project
- project GenAI-GDELT