There is a growing need to strengthen the connection between rapidly expanding information on emerging issues and credible policy decision-making in an increasingly complex and uncertain environment. Traditional horizon scanning approaches face challenges in managing large volumes of information, much of which originate from diverse and often fringe sources, making it difficult to identify, assess and prioritise emerging issues. By exploring the integration of AI into horizon scanning, this paper contributes to improving the evidence base and anticipatory capacity of policy processes. A conceptual framework was developed to examine how AI can support the synthesis of evidence on emerging issues while retaining the critical role of expert judgement. The framework proposes a hybrid methodology that combines AI-enabled automation with human expertise across four interconnected stages: (1) automated scanning of diverse and continuously updated information sources; (2) semi-automated scenario development that synthesises trends, drivers and weak signals; (3) prioritisation of scenarios through combined quantitative and qualitative assessments aligned with policy objectives; and (4) targeted deep dive analysis that translates scanning outputs into actionable policy insights. The framework demonstrates how AI can support more systematic, transparent and scalable approaches to horizon scanning. At the same time, it highlights the continued importance of human expertise for interpreting uncertainty, incorporating contextual knowledge and ensuring transparency, legitimacy and accountability in policy decision-making. Rather than advocating full automation, we argue that effective horizon scanning requires ’human-AI collaboration’ that combines computational capabilities with expert judgement. The framework advances the application of AI within foresight and supports more effective integration of evidence into policy processes. It provides a foundation for developing more transparent, adaptive and evidence-informed horizon scanning and highlights priorities for future research on human-AI collaboration in public sector foresight processes.