In 2024, we made remarkable strides in leveraging AI to combat child sexual abuse. Our flagship initiative, Stella Polaris, has become a cornerstone in our efforts to integrate AI into child protection strategies. By bringing together AI experts, industry stakeholders, researchers, and law enforcement, we have developed cutting-edge tools that enhance the detection and investigation of child sexual abuse material. Key achievements in 2024 were:
Stella Polaris Summit
In June 2024 we hosted the Stella Polaris Summit in Stockholm, a groundbreaking event gathering international experts, leading corporations, and selected guests from various fields to discuss the crucial role of AI in preventing child sexual abuse. The summit fostered interdisciplinary collaborations and showcased innovative AI solutions.
One of the many positive outcomes of the Stella Polaris Summit is that it sparked engaging, at times provoking and controversial discussions. It will be particularly interesting to delve even deeper into these topics in the future, ensuring we are prepared to combat child sexual abuse through cross-sectional collaboration. The summit also highlighted significant gaps between technical and legal perspectives, presenting a unique opportunity for follow-up. This can drive the conversation further, addressing the inflexibility of legislation and fostering much-needed cooperation between the legal and technical domains.
Dan Sexton, CTO Internet Watch Foundation, about the Stella Polaris Summit.
Research convening together with Oxford Internet Institute
A core goal of Stella Polaris is to expand research on AI and child sexual abuse and to encourage PhD students and research programs to engage in this field. Together with the Oxford Internet Institute, we hosted research convening to highlight gaps in current knowledge, strengthen research networks, and contribute to a joint global research agenda aimed at combatting child sexual abuse and exploitation. The event also brought in Swedish researchers and potential funders, expanding the scope to include AI for perpetration prevention—a critical yet often overlooked area.
Strengthening AI tool for chat analysis
In 2023, we funded a pilot project to develop Fronta, an AI tool that analyzes content from five Swedish youth support chats—identifying trends by analyzing topics and words to detect new risks that children are subjected to. The tool will both reduce administrative work, and enhance support for children. In 2024, we funded the next step: integrating Fronta into chat operations, including user training and long-term management. As open-source, Fronta can also be shared with other organizations.
Sharing insights at the We Protect Global Alliance
In December, we were invited to present at the WeProtect Global Alliance Global Summit in Abu Dhabi. Bringing together private companies, governments, civil society, and academia, the summit focused on creating a safer digital world free from child sexual abuse. We shared insights on how AI and cross-disciplinary innovation are helping address this challenge—from analyzing behavior on the darknet to supporting law enforcement and expanding online support services.
Tech to for streamlining investigations
Together with the Swedish Police and international partners, we backed a global hackathon to develop scalable technical solutions that support law enforcement in investigating and preventing online child sexual abuse and exploitation. The Interpol DevOps Hackathon brought together 20 participants from 9 countries and 16 organizations, united by one goal: to disrupt and reduce the harm caused by child sexual abuse and exploitation.
Police officers often spend countless hours reviewing child sexual abuse material in painful detail—work that is both time-consuming and mentally exhausting. This collaboration aims to reduce toxic exposure and streamline investigations, so officers can focus on identifying unknown children instead of documenting abuse.
Early trials showed promising results: improved processing efficiency and reduced exposure. This frees up critical resources, enabling law enforcement to focus on identifying victims and perpetrators more quickly. We´re looking forward to further development and impact of this tool.
This project offers enormous potential, especially in protecting the mental health of investigators by reducing their exposure to distressing material. AI can help relieve this burden, allowing investigators to focus more on prevention and victim identification.
Christian Wallfors, Case Officer at Swedish Police about the DevOps Hackaton
Boosting global detection efforts
A key challenge in detecting and removing child sexual abuse material is that it is illegal to even view or handle the material. In 2019, Childhood hosted an event to address this challenge and identified the need for new tools. In partnership with us, Web IQ developed LIBRA—an AI-driven tool that scans the internet to detect, block, and remove child sexual abuse material. LIBRA identifies both known and likely illegal content without human intervention, making it safer and more scalable for law enforcement and partners.
Childhood has been a key driver of LIBRA’s development—bringing stakeholders together, providing financial and strategic support, and ensuring its legal and scalable use.
Since its full launch in 2022, LIBRA has exposed major child sexual abuse material distribution networks, including three new hotspots containing tens of thousands of images that have now been taken down by police. LIBRA is currently used by 26 law enforcement agencies, NGOs, and trust & safety companies. In 2024, LIBRA was made available to all 53 INHOPE hotlines, significantly boosting global detection efforts.
