One of the most popular apps in Ireland right now is 'SimSimi'. At its most fundamental, the app is a messaging tool that is being used by school children around the country. School principals are now warning parents about the app that is being linked to cyberbullying. 

The app is free to download and no account is needed to use it. This means those who use it are anonymous. Users can type messages into the app and the chatbot will reply. Curious to find out more, I downloaded the app to started to chat to the bot. 

This was the reply I received, with no coaching or influencing on my part. I went to the settings and deselected the 'bad words' option and tried again. 

 

It was at this point that I gave up on SimSimi. 

How does it work?

SimSimi uses artificial intelligence to converse with users. The app learns from previous conversations in an attempt to become more useful to the user. This type of technology is being used by businesses all around the world to help with customer service issues. Technology firms are investing heavily in this area. 

While the concept for this technology is good, it can be manipulated by users, which appears to be the case with SimSimi, and was the case with Microsoft's AI Twitter account, Tay. The chatbot can be taught words, terms and answers by users. It then uses that information when answering future questions. When manipulated, it can become a hate-spouting machine, as appears to be the case here. 

It is possible to flag responses as being 'Not interesting', 'Sexually explicit', Vulgar or violent' or 'Other'. It is not clear as to what happens once an item is flagged. 

The app clearly states "Harass, abuse, defame or otherwise infringe on any other party, you may be subject to civil or criminal penalties.' Again, it is not clear how this is policed. 

We have contacted SimSimi for comment.