CDU Uses DeepSeek & ChatGPT To Expose Disinformation

Academic uses Ai to help fight disinformation-fueled radicalisation.

Charles Darwin University forensic linguistics expert Dr Awni Etaywe has tested whether Ai tools like DeepSeek and ChatGPT can help detect and counter disinformation tactics used in extremist content.

Research Overview

Dr Etaywe, a lecturer in linguistics and a forensic linguistics researcher, examined how DeepSeek and ChatGPT analyse disinformation in far-right extremist material. He fed sections of a white supremacist manifesto into both platforms, comparing their ability to flag misleading tactics.

ChatGPT and DeepSeek provided different yet complementary analyses, focusing on structural, rhetorical, and emotional elements of extremist discourse.

Analytical Differences Between Platforms

According to Dr Etaywe, ChatGPT offered a detailed breakdown of rhetorical techniques, including false authority, exaggerated statistics, and selective presentation of facts.

“ChatGPT provides a broad analysis, helping users understand the structural and rhetorical components of extremist discourse. It breaks down how false narratives are constructed and highlights manipulative framing that contributes to radicalisation,” he said.

DeepSeek, on the other hand, excelled at detecting the emotional and psychological strategies that drive radicalisation.

“DeepSeek focuses on emotional appeals and the divisive nature of extremist content. It identifies how fabricated stories, cherry-picked evidence, and inflammatory language are used to manipulate fear, identity, and urgency,” Dr Etaywe said.

Identifying Disinformation Strategies

Dr Etaywe said ChatGPT was effective at recognising techniques such as distortion of demographic trends, misrepresentation of immigration, and misuse of historical figures’ quotes.

“It flagged false framing, including how Wikipedia is used to give legitimacy to baseless claims,” he said.

DeepSeek, while also detecting disinformation, placed more emphasis on the emotional pull of extremist messaging.

“DeepSeek highlighted how extremist views are reinforced through storytelling, selective evidence, and language designed to provoke strong emotional responses,” he said.

Implications for Countering Extremism

Dr Etaywe said both Ai models could help researchers, organisations, and policymakers develop more informed strategies to combat radicalisation.

“Together, DeepSeek and ChatGPT not only support debunking efforts but also empower individuals, organisations, and governments to develop counter-narratives that undermine extremist ideologies,” he said.

He added that both platforms contribute to broader Ai-driven solutions designed to preserve truth, strengthen social cohesion, and address digital disinformation.

Last Updated on January 31, 2025 by Nick Ross