Gdynia firm’s ‘Cyber Guardian’ leading the way in combatting online violence
A start-up from Gdynia is behind a revolutionary new AI technology which has created the world’s first fully autonomous cyber guardian to detect and intervene in messages of cyber-violence, bullying, paedophile activity and suicidal intentions.
Based on the so-called third wave of artificial intelligence, the innovative combination of both machine learning and reasoning distinguishes Samurai Lab’s technology from existing competition and allows it to independently distinguish genuine cyber-violence with far greater accuracy than others.
Co-founder and CEO Michał Wroczyński explained to TFN that the problem with existing tools for uncovering cyberattacks, is that are very over-reactive. In 50 out of 100 cases, they raise a false alarm because they don’t understand the context of conversations and only see a few key words in the sentence, so they block the whole conversation.
The Samurai Labs team approached the building of their tool in an old-school way. They started from learning grammatical constructions and put the entire knowledge that can be found in an English language textbook, into a computer.
“We added knowledge and reasoning to machine learning creating a neurosymbolic AI which doesn’t just learn based on data, but can also reasons, based on the knowledge it possesses.”
CEO Michał Wroczyński said: “In the communities where we work, we have observed that after a month or two, there is a significant reduction in the level of cyberaggression, by around 45 percent and 75 percent of users who received a message from Samurai, stop attacking others.”
Rather than just registering the incident and waiting for a human moderator to verify whether messages require intervention, Samurai Lab’s cyber guardians respond independently and immediately to untoward, suspicious or worrying messages via an automated bot.
In so doing, it can prevent toxic comments and abuse from escalating in real time while the situation is still evolving, and not after events have taken place.
The technology can be invited and integrated into social networks, gaming communities and entire system networks, such as those used in schools, and the automated bot responses can take different forms, even taking on the guise of a non-playing character in a game.
In school, it is also intended to be a teacher’s assistant, helping to catch instances of bullying in areas when the teacher can’t always be there to monitor the situation.
Rather than just registering the incident and waiting for a human moderator to verify whether messages require intervention, Samurai Lab’s cyber guardians respond independently and immediately to untoward, suspicious or worrying messages via an automated bot.
Wroczyński said: “We’ve manged to prove that the technology has been able to reduce the level of violence in some really dark corners of the internet like Men’s Rights on Reddit by 20%."
In the communities where the technology is present, the team have observed that after a month or two, there is a significant reduction in the level of cyberaggression, by around 45 percent, and 75 percent of users who received a message from Samurai, stop attacking others.
The company is now working on a UK-based project called the Hate Lab, in collaboration with Cardiff University where Samurai Lab’s technology monitors the behaviour of online users of the British internet with regard to cyber-violence directed at Poles, distinguishing 11 different categories of aggression.
Wroczyński told TFN: “We would really like to get this technology into the Polish school networks. Poland is quite unique in that all schools in Poland are part of the same network, which would allow for the technology to be implemented quite easily across all schools, we are hoping to make that happen.”
