10101274AI-powered narrative extraction and classification to combat misinformation and disinformationClosedCollaborative R&DInnovate UKIn 2020, when Valent Projects began operations, disinformation and online manipulation were almost exclusively political problems. Since then, the same techniques have been used to manipulate financial markets and sabotage large corporations. At Valent, we detect and track the activity of inauthentic networks that drive disinformation and online manipulation for government and private sector clients. Bad actors are using social media as a way to manipulate narratives, enabling what were once minority voices to appear much louder and more prevalent. Public policy related to transport (particularly when intersecting with climate issues) and the construction/infrastructure sectors are frequent targets of such manipulation. Valent's recently published investigations have resulted in commercial interest. As manipulation techniques have increased in scale and complexity, the technology to combat them has had to evolve. AI and machine learning are crucial to transforming our existing, labour-intensive methodologies into scalable solutions capable of meeting rising demand from clients and potential clients facing ever-more complex threats. Ariadne, our bespoke software tool has been developed to automate the detection and tracking of inauthentic online networks. This grant will enhance Ariadne's capabilities by allowing for the analysis of strategic narratives. We will implement machine learning techniques to detect and classify strategic narratives to sustainably scale. AI assurance will also be implemented for reliable and trustworthy outcomes. The focus in this project is understanding audience sentiment to gauge the influence of strategic narratives, especially those stemming from an inauthentic source.