AI for predicting particularly serious offences?

Cases of sexual exploitation of children are increasing every year. In 2021, 29.3 million reports of suspected sexual exploitation of children were received by protection organisations in the USA alone. Worldwide, up to 35 per cent of girls and 21 per cent of boys experience sexual violence before their 18th birthday. At any given time, an estimated 750,000 people are seeking contact with children online for sexual purposes, posing a major threat to the 800 million children who actively use social media. The number of users who consume child sexual exploitation material is much higher. The numbers are rising steadily, especially because the majority of activities take place on the dark web, i.e. completely anonymously and hardly traceable. The sheer number of offences and consumers and the obscuring effect of the dark web pose a major challenge for law enforcement authorities worldwide: the limited resources available should be used primarily where particularly serious offences are likely to occur. But how can you filter out those at high risk of sexually exploiting children from the millions of users on the dark web?

Professor Thomas Schäfer, holder of the Professorship for Quantitative Methods in Psychology at the Health & Medical University Erfurt, will present an EU-funded project in which artificial intelligence was used to determine this risk. Data from many dark web forums was paired with already known scientific findings to develop algorithms to predict the risk of serious criminal offences. The result of the project is a programme that can be used by law enforcement agencies worldwide to identify high-risk offenders.

The other dates of the AI lecture series in the summer semester can be found on our event poster (in German only).