There are many tools that exist for parental surveillance of a child's social media accounts, so why are cyber-threats to children still a problem? We talked to professionals that deal with online dangers and the answer is pretty clear - the current tools do not adequately address the problem. How do you keep track of the huge level of social media activity, activity across multiple social media accounts, and identify more sinister threats that are not immediately recognisable? More importantly, a child needs to be able to access social media. Not only are children desperate find their way around parental surveillance tools, but the best way for a child to learn how to use social media safely is by being exposed to it. Parenting resources do not actually recommend the use of these tools due to the trust issues they cause.

The product

The frustrating part of this problem is that compared to real life threats, all of the information and control is there at our tips of our fingers and we can do nothing. That is why we developed the Bufferlo Cyber Safety platform, a cyber safety tool that uses artificial intelligence to monitor and detect cyber threats at a level never before possible, while still allowing children to use social media privately. The key to its success is our approach to cyber safety. Messages and images from the child's social media are analysed by our state of the art AI technology. Because we can reliably detect threats, the parent does not have to monitor the child's social media, but is alerted as soon as threats are detected.

What it does

In designing the AI technology, we needed to look at what would be the most effective way to prevent online dangers. For our online predator detection we looked at message data of real chid predators, and trained an artificial intelligence on that data. Using recent advances in artificial intelligence, our neural networks are able to analyse complex patterns of text to measure intent regardless of the language used. Our AI does not just look for keywords in text - it's an expert in the way child predators act and the techniques they use to target victims. In a test scenario, 81.75% percent of child predators were detected within the first 25 messages. The same technology is currently being developed to provide protection against other threats, including cyber-bullying and drug and alcohol solicitation. For our nudity detection, developing our technology to have leading performance was just the first step. We developed technology with the intelligence to detect the specific region within an image that is sexually explicit, allowing just those particular regions to be filtered. We also added facial recognition technology for detecting if someone's child is showing up in a sexually explicit image. Facial recognition of children is traditionally a problematic task, due to the abnormal variation in age and other features such as hair style and eyewear. Our facial recognition technology uses new technology to solve these problems with state of the art results.

What's next for Bufferlo Cyber Safety

The Bufferlo cyber safety platform is almost ready for public access and will launch later this year, offering a subscription platform to parents on web, iOS, and Android. The sophistication of our technology and the approach of our platform are game changing for cyber safety. Never before has a child had the privacy freedom to use social media while still being kept safe with technology that will detect child predators, will reduce cyber-bullying, and will reduce the spread of child pornography.

Share this project: