On December 12 and 13, the INHOPE network came together for an online Hotline Training Meeting to discuss Artificial Intelligence. Implemented on behalf of the European Commission, the goal of these two days of meetings was to provide insights into Artificial Intelligence and the potential opportunities and challenges it poses for our network and the broader child protection ecosystem.


The European Commission provided hotlines with a comprehensive introduction to Artificial Intelligence (AI) and Large Language Models (LLMs). This session equipped hotlines with a fundamental understanding of AI and LLMs, how they function, and their broader applications. This foundational knowledge is crucial for comprehending the significance of AI in the context of addressing child sexual abuse material (CSAM) and its related challenges.

PWC and Offlimits presented an AI tool that will enhance the prioritisation of CSAM-related cases within the child protection ecosystem. This tool, currently in phase 2 of testing, will empower hotlines to efficiently prioritise CSAM cases and allocate resources more effectively. Europol provided insights into the evolving nature of online crimes and how law enforcement agencies are tackling the challenges posed by artificial intelligence. While the risks of AI were discussed in great detail, the session ended by emphasizing that AI can assist LEA and hotlines greatly by providing information and guidance to help accelerate investigation procedures.

GCHQ then demonstrated its Site Fingerprinting tool, which helps automate the discovery and identification of CSAM sites and helps to identify recurrences of the same site. Hotlines engaged in a discussion on how to best engage with this tool as a network and will explore future avenues of cooperation as this tool will be a powerful resource for combating CSAM effectively.

To close the two-day discussion, Open.AI presented the industry's approach, particularly that of generative AI model builders, concerning safety design for online child exploitation and abuse in connection to generative AI models. The primary goal was to establish a shared understanding of high-level current safety and responsible AI practices at different levels of the tech stack, alongside a broad overview of existing Online Child Sexual Exploitation and Abuse (OCSEA) Risk Mitigation practices in the industry.