Fighting Multimodal Misinformation
Malicious actors use generative AI to create misinformation to deceive, manipulate, and spread harmful narratives at scale. In response, this area develops methods, benchmarks, and workflows for the detection, contextualization, and moderation of misleading or machine-generated content, contributing to the broader vision of ethical, robust, and socially grounded NLP. To combat the proliferation of fake or misleading claims, this area advances general-purpose fact-checking systems that operate reliably across domains. Benchmarks and models are developed to reflect real-world scenarios, support human-interpretable outputs, and align with journalistic standards.