Captchat: A messaging tool to frustrate ubiquitous surveillance

Paul Dunphy, Johannes Schöning, James Nicholson, Patrick Olivier

Research output: Contribution to conferencePaper


There is currently a widespread uncertainty regarding the ability of citizens to control privacy online in the face of ubiquitous surveillance. This is a huge and complex societal problem. Despite the multi-faceted nature of the problem, we propose that HCI researchers can still make a positive contribution in this space through the design of technologies that support citizens to engage with issues of surveillance. In this paper we describe the design of a messaging application called Captchat. Captchat enables people to send everyday messages embedded into images, with the added ability to apply visual distortions to the message to resemble an online CAPTCHA. We propose the chief benefit would be that Captchat messages (with potentially "one-time" distortions) can increase the difficulty for algorithms to index private messages and necessitate the involvement of much more costly human labor in the surveillance process. We developed a prototype and conducted a user study; the results suggest that people were likely to create Captchat messages that were difficult to index for an OCR package but still easy to understand by humans, even without explicit instructions to interact 'securely' with the application. While more work is still required to understand the limitations of Captchat, we hope it can open discussion on how HCI researchers can respond to the challenges faced from ubiquitous surveillance.
Original languageEnglish
Publication statusPublished - 18 Apr 2015
Event33rd Annual ACM Conference Extended Abstracts on Human Factors in Computing Systems - Seoul, Korea
Duration: 18 Apr 2015 → …


Conference33rd Annual ACM Conference Extended Abstracts on Human Factors in Computing Systems
Period18/04/15 → …


Dive into the research topics of 'Captchat: A messaging tool to frustrate ubiquitous surveillance'. Together they form a unique fingerprint.

Cite this