Abstract
This article presents a taxonomy of incentive vulnerabilities that can affect public and consortium blockchain-based networked intelligent systems. The taxonomy aims to help researchers and developers better understand the related threats and design more secure systems. To this end, the proposed taxonomy is grounded in a generic multi-agent organizational model for blockchain systems (AGR4BS) and establishes a relationship between the vulnerabilities and the dedicated agent roles. We expressed the vulnerabilities as behavior deviations and classified them according to the roles and behaviors identified in AGR4BS to form the categories and refine the subcategories of the taxonomy. The proposed taxonomy is novel and distinctively different from other taxonomies found in the literature.