Three researchers from Stanford University have found ways to captchas, which, inter alia, spambots should go, to crack. The security of large sites such as eBay and Wikipedia it turns out that there easy to pass.
A captcha displays a number of letters and figures in such a way verhaspeld that they are known by people but not by computers can be read. That would be the spambots make it impossible for automated accounts to register; registratieformuleren on a lot of websites make use of captchas.
With the Decaptcha project are researchers from the Stanford University have managed to develop ways to make the automatic websitebeveiliging to crack. The program removes the visible ‘noise’ in the images, put the letters in plain text, and makes the captchas for computers to read. This happens in a fraction of a second, in the research.
The researchers tried their software out on several large sites, with various types of captchas. Large sites like eBay, Wikipedia, CNN, Baidu, Digg and Visa pages Authorize.net proved to work. Of the fifteen tested captchas remained only Google and Recaptcha reasonably out of the danger zone. Their captchas were other less often cracked because of the methods used, the more complicated the number and the average size of the characters to determine.
Previously showed the Stanford computer scientists all vulnerabilities in the audio captcha’s. The layers to the way the software works; the figures and letters in the captchas that are in the audio clips are audible, form peaks in a spectral analysis. Those peaks of noise are separated, but in speech as a disturbing component it will not work. Audio captchas are used to visually impaired people to help solve a captcha.
