20 lines of text is all that is needed to detect a potential online predator. The signing of a formal agreement between the Norwegian Police and Aiba will further boost the precision in detection of online abuse against children.

“There is no doubt that the police are most useful when preventing crime,” says police inspector Rune Otterstad in the Innlandet Police District.

On Wednesday, the Innlandet Police District entered into an agreement with Aiba.

Through real-time communication analysis, Aibas AI will become better at preventing sexual abuse online. The agreement has been approved by the Director of Public Prosecutions.

Learning from the police
Prevention is not something the police can do alone, the police inspector pointed out. “The most important thing is to prevent children from being exposed to abuse. Because once that happens, it’s too late, and we often see that victims of this type of crime struggle for many years,” says Otterstad.

By providing the AI with insights into how the police work in online abuse cases and what kind of behavior they look for in current communication, the AI is trained to detect conversations that could lead to sexual abuse.

The police have been collaborating with Aiba since the company was a project at NTNU Gjøvik with the approval of the Director of Public Prosecutions. The agreement is a formalization of the collaboration.

Preventing grooming initially, the robot will not be used by the police themselves, but by gaming companies and companies involved in social media.

By monitoring communication in real-time and ranking the content based on a risk score:  The higher the score, the greater the risk of potential grooming. The idea is that the gaming companies’ own moderators or security teams can intervene and stop further conversations before harm occurs.

Preventing harm before it happens
“The goal is to ensure that all children and young people are safe online, not just in Norway, but worldwide,” says Hege Tokerud, CEO of Aiba.

In March of this year, the company secured the online game Moviestar Planet as a customer.

According to Tokerud, the Danish company has 100 million children on its platforms worldwide. The AI is active on the American servers, where it monitors about one million chat rooms.

Several servers and other potential gaming companies are pending, she says.

Multiple potential purposes
Although the goal is to prevent grooming, the AI can also be used to detect bullying, self-harm, and radicalization, for example.

Both Tokerud and Otterstad in the police believe it’s significant to be able to address problems instead of waiting until the damage has occurred.

“It will be a great relief for us to have fewer of these cases,” says Otterstad.