Net Security Purr-Fected: Pictures Of Kittens Are The Unlikely New Weapon Against Online Fraud And Spam

There’s a new way to combat internet fraud, prevent spam and keep online shopping secure. But your first impressions may be that it’s not exactly high tech. It takes the form of a simple question: From a gallery of fluffy-animal snaps, can you tell which are cats and which are dogs?

Your answer is enough to find out whether you are human or an automated spam program, designed to send unwanted email.

The dog/cat question is the latest example of a security device called a Captcha, a simple puzzle that usually takes the form of a string of distorted letters and numbers.

Captcha stands for Completely Automated Public Turing Test to Tell Computers and Humans Apart.

The idea behind a Captcha is that users have to perform a task that is simple for a human but incredibly difficult for a computer. Distorting random letters and numbers makes them confusing to a computer but readable to the human eye.

Regular web users will be familiar with Captchas, as they are ubiquitous on shopping, email and networking sites; during initial registration and sometimes log-in, Captchas are used as an additional gateway to passwords.

Although a number of computer researchers have claimed that they invented the Captcha, it’s generally acknowledged that Carnegie Mellon University led the charge after being asked by Yahoo in 2000 to create a security tool to stop spammers using computer programs to set email accounts and then use these accounts to send millions of spam messages.

According to Luis von Ahn, a member of the original Carnegie Mellon team, “Captchas are still the best defence against many types of automated attacks, and I believe they will be used for the foreseeable future. The only ones that can be broken are the extremely primitive ones that use a constant font, and apply no distortion to the characters other than thin lines that are easy to remove automatically.”

But as programs are written that can read heavily distorted codes, the distortions become even more extreme. And as they do so, some of the Captchas are becoming too tricky for many humans to decipher at first attempt. More and more users are finding that they need two or three attempts before they can confirm their shopping orders or set up their new email account. So, creators of Captchas are exploring new avenues.

Von Ahn is the executive producer of a new project,, which uses old tomes to create new Captchas. While digitally scanning books to make them available online, character recognition software often fails to recognise a word, because of smudges or damaged paper. If von Ahn’s software can’t read it, he’s assuming that other computers will also struggle. “The words in my Captchas come directly from old books that were recently scanned, and we are using people’s answers to decipher what the words are.”

Picture recognition is an increasingly popular alternative. People are asked to look at a grid of images and pick the ones that have something in common – straightforward for humans but impossible for computers, as it’s difficult for computers to accurately classify images.

Pix Captcha (, a Carnegie Mellon project, displays pictures of certain things – worms, babies and so on – and then asks people to select the corresponding noun from a drop-down menu.

Most altruistic is a Microsoft research project called Asirra ( – Animal Species Recognition for Restricting Access – that uses pictures of rescue-home dogs and cats from It asks you to click on the cats, rather than the shots of aardvarks, bears and dogs thrown in to baffle the computers.

It also helps find homes for domestic animals – each image has a tag reading “adopt me” on it.

Although still in the “beta” testing stage, Asirra has a database of over two million images with which it can create Captchas. It has the potential to change the way we stay secure online – and give animal lovers everywhere a dose of cuteness.

Adapted from The London Independent.