This article addresses the increasing spread of misinformation and disinformation through deepfakes and AI-generated content. Such content poses a serious threat to societal stability and was classified as a critical short-term risk by the Global Risks Report 2024. Effective fact-checking is essential, but is limited by the speed of digital dissemination. The article examines regulatory requirements for the use of AI and presents current research projects on AI-supported disinformation verification the BMBF-funded collaborative project newspolygraph, which is developing a modular, multimodal technology platform for the detection and analysis of disinformation across text, audio, image, and video content; the BMBF-funded junior research group project VeraXtract, which is dedicated to researching and visualizing disinformation narratives at various levels of abstraction and creating an interactive platform for political education; and the BIFOLD agility project FakeXplain, which explores personalized explanations for AI systems used for disinformation detection. These projects combine technological innovation with ethical and legal requirements to improve human-machine interaction, promote trust, and contribute to democratic resilience.