Deepfakes are an emerging communication innovation possessing vast implications for innovation scholarship. A systematic literature review of multidisciplinary literature was undertaken to assess existing deepfake definitions and synthesize a new definition to guide future theoretical development and empirical understanding of this communication innovation. Further, the systematic review identifies deepfake creators and those depicted by deepfakes and evaluates value creation and destruction implications of deepfakes for customers and organizations. Following the PRISMA protocol, this review evaluates deepfake research published between January 2017 and June 2021 across five databases, including only English literature from Q1/Q2 peer-reviewed journals. Eighty research articles were included in the final review. Using text mining software, a new deepfake definition is synthesized which encompasses the emerging concepts of “videos”, “audio”, “realistic”, “fake”, “artificial”, “learning”, “media”, and “saying”. Undetermined actors and individual content creators were most commonly identified as deepfake creators, whereas public figures, celebrities, and actors were most frequently depicted by deepfakes. Deepfakes potentially create and destroy value for customers and organizations. This study provides a new, holistic multidisciplinary definition of deepfakes, offers fresh insights into the use and impact of deepfakes as a communication innovation, and provides a new understanding of the value implications derived from deepfakes for innovation. Lastly, a future deepfake research agenda for innovation scholars is provided. •A new deepfake definition integrating multidisciplinary perspectives is synthesized.•Seven groups who create deepfakes are identified.•Twelve groups depicted by deepfakes are identified.•The value implications of deepfakes for customers and organizations are examined.•A research agenda guiding future innovation research into deepfakes is proposed.