Unisys researches use of AI, ML to detect deceitful writing

Unisys has received research funding from the Artificial Intelligence (AI) for Decision Making Initiative to investigate the use of AI and machine learning (ML) to detect deceitful and persuasive writing.

The AI for Decision Making Initiative is a collaboration between the Office of National Intelligence (ONI) and Department of Defence to develop AI and ML expertise in areas of importance to the Australian Defence Force and national security community.

Of the 23 challenges set by the initiative, Unisys applied to develop an AI- and ML-based solution to detect deceitful or persuasive writing in a collection of documents by the same author.

Dr Xin Xu, practice lead for Artificial Intelligence, Machine Learning & Data Science for Unisys Asia Pacific, said the company ultimately wants to be able to detect if an author’s writing style has changed in documents making exaggerated claims.

“Real-world applications could potentially include forensic analysis to determine who is the true author of a written piece for legal purposes; identify if emails or text messages are written by imposters such as hackers or phishers; or even detect fraudulent or fake statements on Twitter or other public information platforms designed to confuse media or the public,” Dr Xu said.

With the research funding awarded in November 2020, Unisys will experiment, prototype and demonstrate the proposed approach using publicly available data in a technical paper and oral presentation in the first quarter of 2021.

Image credit: ©stock.adobe.com/au/Grispb