Abstract: Previous studies in psychology and linguistics revealed that people who committed fraud in online chats would show certain cues in their text/words. For example, deceptive senders displayed higher non-immediacy and expressiveness; used more words but in less complexity and less use of self-reference and diversity in their messages compared to truthful senders. Nevertheless, previous researchers did not explain how much was deemed to be less or more, compared to truthful senders. Such previous studies only determined whether a message was deceiving or not. Therefore, in order to detect deceptive senders, a deception detection model has been developed. It was integrated into a newly developed chatting software to analyse the degree of trust in chatting partners. An experiment was conducted to test the level of user trust in the developed software. The results of the experiment indicated that the software analysis produced an 11.36% possibility that a participant has been deceived. Although there was only a 11.36% chance that a chatting partner had been deceived, this proved that fraud had been committed by the chat partner. The experiments further found that the participants believed in the given results of the analysis and this affected their decision.