Vera Schmitt is Professor at Johannes Gutenberg University Mainz and founder and head of the XplaiNLP group, which is situated at JGU Mainz and assoziated with TU Berlin, DFKI and CERTAIN. Her research focuses on trusted information processing with language models: developing methods that make AI systems explainable, robust, and reliable, so that people can understand, verify, and appropriately rely on what these systems produce. Her group combines interpretability and explainability (attribution methods, natural-language rationales, counterfactuals, and analysis of model internals), model steering toward more reliable and less biased outputs, reliable evidence retrieval, and human-centered evaluation of how explanations and model outputs affect trust, decision quality, and task performance. These technical contributions are connected to regulatory frameworks such as the EU AI Act and the Digital Services Act.

She pursues this agenda through two central use cases. The first is clinical decision support, where her group studies how language models behave in medical settings, including biases in their outputs, and investigates how their outputs can support clinicians in their decisions (e.g., in the project like XPLAINER, VERANDA). The second is disinformation and information manipulation, where her group develops methods for (semi-)automated fact-checking, evidence retrieval, narrative monitoring, and the detection of disinformation, deepfakes, and coordinated manipulation campaigns (e.g., in the projects FIMI-RESIST, VeraXtract, and FakeXplain). She coordinates the EU Horizon project FIMI-RESIST, and further projects include XPLAINER and the DFG-funded ReDis project.

She has raised over €6.5 million in third-party funding, is a GI Junior Fellow, and is active in the community as Area Chair, reviewer, and keynote speaker.