Worried about the robustness of the ubiquitous CAPTCHA in preventing online attacks? TRUST researchers Jeremiah Blocki, Manuel Blum, and Anupam Datta at Carnegie Mellon University think they have the solution: GOTCHAs, or Generating panOptic Turing Tests to Tell Computers and Humans Apart. GOTCHAs introduce a new twist on the use of human-only solvable puzzles (such as CAPTCHAs) by having the interaction between humans and computers use Inkblot images and associated phrases used to describe the images. A user is presented with an image and asked to describe that image with a short phrase. Later, when logging in online the user is shown the same image and challenged to provide the matching phrase that describes it. For a human, successfully completing both steps is easy. For a computer, not so much. Hence GOTCHA's effectiveness.

See MIT Technology Review and Gizmodo Australia for coverage of this research. The full research paper is available here.