A CAPTCHA (an acronym for "completely automated public Turing test to tell computers and humans apart", trademarked by Carnegie Mellon University) is a type of challenge-response test used in computing to determine whether or not the user is human. The term was coined in 2000 by Luis von Ahn, Manuel Blum, Nicholas J. Hopper of Carnegie Mellon University, and John Langford of IBM. A common type of CAPTCHA requires that the user type the letters of a distorted image, sometimes with the addition of an obscured sequence of letters or digits that appears on the screen. Because the test is administered by a computer, in contrast to the standard Turing test that is administered by a human, a CAPTCHA is sometimes described as a reverse Turing test. This term, however, is ambiguous because it could also mean a Turing test in which the participants are both attempting to prove they are the computer....
Since the early days of the Internet, users have wanted to make text illegible to computers. The first such people were hackers, posting about sensitive topics to online forums they thought were being automatically monitored for keywords. To circumvent such filters, they would replace a word with look-alike characters. HELLO could become
)-(3££0, as well as numerous other variants, such that a filter could not possibly detect all of them. This later became known as "13375p34k" (leetspeak).
The first discussion of automated tests which distinguish humans from computers for the purpose of controlling access to web services appears in a 1996 manuscript of Moni Naor from the Weizmann Institute of Science, entitled "Verification of a human in the loop, or Identification via the Turing Test". Primitive CAPTCHAs seem to have been later developed in 1997 at AltaVista by Andrei Broder and his colleagues in order to prevent bots from adding URLs to their search engine. Looking for a way to make their images resistant to OCR attack, the team looked at the manual to their Brother scanner, which had recommendations for improving OCR's results (similar typefaces, plain backgrounds, etc.). The team created puzzles by attempting to simulate what the manual claimed would cause bad OCR recognition. In 2000, von Ahn and Blum developed and publicized the notion of a CAPTCHA, which included any program that can distinguish humans from computers. They invented multiple examples of CAPTCHAs, including the first CAPTCHAs to be widely used (at Yahoo!).
CAPTCHAs based on reading text — or other visual-perception tasks — prevent visually impaired users from accessing the protected resource. However, CAPTCHAs do not have to be visual. Any hard artificial intelligence problem, such as speech recognition, can be used as the basis of a CAPTCHA. Some implementations of CAPTCHAs permit users to opt for an audio CAPTCHA.
The development of audio CAPTCHAs appears to have lagged behind that of visual CAPTCHAs, however, and presently may not be as effective. Other kinds of challenges, such as those that require understanding the meaning of some text (e.g., a logic puzzle, trivia question, or instructions on how to create a password) can also be used as a CAPTCHA. Again, there is little research into their resistance against countermeasures.
Some interesting tests came on the idea of image recognition. One such example is the KittenAuth, a test that asks the user to recognize some certain animal (kittens) in a series of pictures of multiple species (dolphins, puppies, foxes...)
For non-sighted users (for example blind users, or the color blind on a color-using test), visual CAPTCHAs present serious problems. Because CAPTCHAs are designed to be unreadable by machines, common assistive technology tools such as screen readers cannot interpret them. Since sites may use CAPTCHAs as part of the initial registration process, or even every login, this challenge can completely block access. In certain jurisdictions, site owners could become target of litigation if they are using CAPTCHAs that discriminate against certain people with disabilities. In other cases, those with sight difficulties can choose to identify a word being read to them.
While providing an audio CAPTCHA allows blind users to read the text, it still excludes those who are both visually and hearing impaired.
The use of CAPTCHA thus excludes a large number of individuals from using significant subsets of such common Web-based services as PayPal, GMail, Orkut, Yahoo!, many forum and weblog systems, etc.
Even for perfectly sighted individuals, new generations of CAPTCHAs, designed to overcome sophisticated recognition software, can be very hard or impossible to read. Even some of the demo CAPTCHAs at the software sites listed below are indecipherable to many if not all humans.
There are a few approaches to defeating CAPTCHAs: using cheap human labor to recognize them, exploiting bugs in the implementation that allow the attacker to completely bypass the CAPTCHA, and finally improving character recognition software.
Cheap human labor
It may be possible to subvert CAPTCHAs by relaying them to a sweatshop of human operators who are employed to decode CAPTCHAs. The W3C paper linked below states that such an operator "could easily verify hundreds of them each hour". Nonetheless, some have suggested that this would still not be economically viable. (e.g. ) Paying the human operators with access to pornography instead of money has also been considered.
Some poorly designed CAPTCHA protection systems can be bypassed without using OCR simply by re-using the session ID of a known CAPTCHA image. Sometimes, if part of the software generating the CAPTCHA is client-sided (the validation is done on a server but the text that the user is required to identify is rendered on the client side), then users can modify the client to display the unrendered text, etc.
Computer character recognition
Although CAPTCHAs were originally designed to defeat standard OCR software designed for document scanning, a number of research projects have proven that it is possible to defeat many CAPTCHAs with programs that are specifically tuned for a particular type of CAPTCHA. For CAPTCHAs with distorted letters, the approach typically consists of the following steps:
- Removal of background clutter, for example with color filters and detection of thin lines.
- Segmentation, i.e. splitting the image into segments containing a single letter.
- Identifying the letter for each segment.
Step 1 is typically very easy to do automatically. In 2005, it was shown that neural network algorithms have a lower error rate than humans in step 3. The only part where humans still excel computers is step 2. If the background clutter consists of shapes similar to letter shapes, and the letters are connected by this clutter, the segmentation becomes nearly impossible with current software. Hence, an effective CAPTCHA should focus on step 2, the segmentation.
Neural networks have been used with great success to defeat CAPTCHAs as they generally are indifferent to both affine and non-linear transformations. As they learn by example rather than through explicit coding, with appropriate tools very limited technical knowledge is required to defeat more complex CAPTCHAs.
Some CAPTCHA-defeating projects:
- Mori et al. published a paper in IEEE CVPR'03 detailing a method for defeating one of the most popular CAPTCHAs, EZ-Gimpy, which was tested as being 92% accurate in defeating it. The same method was also shown to defeat the more complex and less-widely deployed Gimpy program 33% of the time. However, the existence of implementations of their algorithm in actual use is indeterminate at this time.
- PWNtcha has made significant progress in defeating commonly used CAPTCHAs, which has contributed to a general migration towards more sophisticated CAPTCHAs.
- A number of Microsoft research papers describe how computer programs and humans cope with varying degrees of distortion.