Tiscali: Child Protection Online

Will Crisp just end up sending me alerts every time my child or their friends use a swear word?

No. The Crisp Anti-Grooming Engine analyses over two hundred different aspects of each conversation in order to assess whether a young person is being groomed or not. Profane and sexual language is only one aspect of that analysis and use of this kind of language alone is not enough to trigger an alert.

How does Crisp avoid alerting me to normal sexual chat between my son/daughter and their girlfriend/boyfriend?
In order to test their engines accuracy, Crisp sent it for independent analysis with a Professor of Applied Statistics at Cambridge University. Completely new data, previously unseen by the engine, was fed through it. Some of that data was from known groomers but most was ordinary teen-chat. Crisp spotted every instance of grooming behaviour in this test. Not only that, it also managed not to create any false alerts (from the normal teen-chat) 98% of the time.

Crisp can do this because it works by comparing the conversations taking place on your home PC with minutely detailed profiles of real-life groomers (who have been convicted of their crimes) and real-life boys and girls (who volunteered to have their chat used for this purpose). It looks at many different facets of each conversation including sexual content and punctuation, aggression levels and sentence length, typing speed and vocabulary. By working in this way the engine is able to understand when a conversation is being manipulated by a groomer and when it involves two teenagers even if those teenagers are boyfriend and girlfriend using sexual language.

More importantly, the profiles on the database are changing constantly so the engine learns as it goes along, rather than matching pre-set patterns. If the online predators change their tactics and behaviour, this is reflected straight away in adjustments to the way Crisp assesses its threat ratings. If youngsters start visiting new social networking sites or using different slang, these changes will quickly modify the Anti-Grooming Engine's idea of what a typical 10-year-old boy or a 14-year-old girl might be like. Over time, the more relationships Crisp analyses, the smarter it gets at detecting the bad ones.

How do I know the service is working if I do not get any alerts from Crisp?
Naturally we hope that you never get an alert from Crisp. But you can be certain that the programme is working properly by checking the following:

Check the monthly reports that Crisp send you and make sure that they have recorded that the number of lines of chat they have analysed during that period is greater than zero,

When you start up your PC make sure that Crisp appears briefly telling you that this computer is protected by Crisp

When your PC is working check that you can see the Crisp icon in the tool bar on the bottom right of the screen

What happens when Crisp informs parents?
Parents instinctively know how best to protect their children. It is Crisps responsibility to alert you to any potential threat. How you deal with the threat once you have read the Crisp alert is up to you. In rare circumstances Crisp might have generated a false alarm in which case you might like to contact Crisp to tell them so that they can fine tune their engines performance even more.

If, on the other hand you feel that the alert Crisp has sent to you is a serious matter then you may want to speak to your child or contact the Child Exploitation and Online Protection Centre (CEOP) in order to get the police involved.

How do Crisp inform parents of suspicious activity?
You can be informed by Crisp either by SMS or email, depending on your preferences. During the sign-up process you will be asked how and where you would like Crisp alerts to be sent to you.