Dis/misinformation is not experienced equally across languages. Of the world's 7,000+ living languages, roughly 90% are considered low-resource; only around one hundred have meaningful representation in the datasets used to train AI systems, and a large majority of online content is produced in a handful of dominant languages, particularly English. The consequence, documented across every strand of evidence reviewed for this project, is a structural protection gap: the languages least served by platform content moderation, fact-checking infrastructure and language technology are also the languages in which false and manipulative content can circulate longest without detection or correction. Speakers of minority, minoritised and medium-sized languages are therefore doubly exposed: first as targets of identity-based disinformation, and second as communities that the traditional media and counter-disinformation ecosystem are least equipped to defend. As was often highlighted during the project activities, generative AI sharpens both edges of this problem. Large language models (LLMs) lower the cost of producing fluent, localised and personalised disinformation in many languages at once, while performing worst in output quality, safety-filter robustness and detection accuracy, precisely in low-resource languages. At the same time, these models offer genuinely new capabilities for defenders, enabling cross-lingual transfer of detection tools to languages for which no dedicated resources exist. Whether LLMs prove friend or foe to minority-language communities depends, in large part, on political and policy choices. Ireland is a revealing case. As an Anglophone country, it imports mis/disinformation from the much larger US and UK information spheres without any translation lag, while Irish (Gaeilge) and significant community languages, such as Polish, sit largely outside the State's counter-disinformation, media-literacy and public-information efforts. Ireland currently lacks a permanent, systematic research infrastructure for monitoring the volume, sources and targets of disinformation across languages. As a result, policymakers lack a robust domestic evidence base on the scale and characteristics of the problem. In order to address this gap, the recommendations below address the European, Irish, research and platform levels.