Abstract
Disinformation is becoming more prevalent in the corporate sphere, especially as brands choose to promote their products through influencers or micro-celebrities who are perceived as reliable and impartial, but may facilitate false information. The spread of disinformation can have negative economic impacts on companies and brands, which can even affect their reputation. Artificial Intelligence can help detect false information and has become
increasingly important in combating disinformation. The current work addresses the problem of characterizing multimodal disinformation targeting corporations and provides a collection of content that spreads disinformation in digital media. The content was manually annotated with information about the target (Organization, Brand,
or Other) and the source (Corporate, Advertising, or Other) of the false content. We conduct comprehensive experiments to evaluate the effectiveness of state-of-the-art Unimodal and Multimodal Large Language Models in identifying the source and target of the content.
