Abstract: Patent applicants write patent specifications that describe embodiments of inventions. Some embodiments are claimed for a patent, while others may be unclaimed due to strategic considerations. Unclaimed embodiments may be extracted by applicants later and claimed in continuing applications to gain advantages over competitors. Despite being essential for corporate intellectual property (IP) strategies, unclaimed embodiment extraction is conducted manually, and little research has been conducted on its automation. This paper presents a novel task of unclaimed embodiment extraction (UEE) and a novel dataset for the task. Our experiments with Transformer-based models demonstrated that the task was challenging as it required conducting natural language inference on patent specifications, which consisted of technical, long, syntactically and semantically involved sentences. We release the dataset and code to foster this new area of research.