Enhancing online protection and accountability with generative AI
Challenge
UNHCR Hungary faces challenges in addressing the increased activity within online communities of Ukrainian refugees. Current manual monitoring methods are resource heavy and prone to errors and language issues. As online-generated content increases, a more efficient approach is needed to engage with voices of refugees in online communities.
Solution
Explore, in collaboration with refugees, ways of using AI to follow online conversations, and gather and analyze information shared by/targeted at forcibly displaced people. Finding ways to use generative AI safely and ethically will help identify patterns and keywords, enabling UNHCR to accurately detect emerging needs and potential threats.
Impact
Enhanced digital engagement and inclusion with refugee communities online, ensuring timely and effective intervention strategies. By working with 150 refugees directly and potentially reaching up to 11,000 more, the project will find ways for UNHCR to gain understanding of refugees’ needs and challenges based on their digital interactions.
Project impact
Other information
Having designed, developed, and tested an AI-powered tool to help monitor online content for early identification of protection needs, the project developed SOPs, a taxonomy, and an ethical labelling framework for responsible, AI-supported monitoring of online content, enabling training and institutional knowledge-sharing. Read more about this initiative here.