<p>News headlines are an important source of information that concisely encapsulates events, trends, and complex stories. However, it can be difficult to extract structured information from these headlines due to their brevity, incompleteness, and use of idiomatic expressions. Although recent studies have explored information extraction from domain-specific biomedical articles’ headlines, the work for open-domain information extraction from news headlines remains limited. This paper introduces a novel approach to extracting the information from news headlines. We aim to find the claims from the headlines and then extract the subject, relation, and object from the news headlines. To accomplish this, we first developed a corpus of Claim News Headlines (CNHeadlines) and classified the headlines as either a claim or not a claim. Next, we created a new framework called Headline Fact Extraction (HFE), which leverages open information extraction to extract subjects, predicates, and objects. Finally, we compare and evaluate different open information extraction (OpenIE) models for extracting structured information from news headlines. The experimental results show that our proposed methodology HFE achieves a 98% <i>F</i>1 score for the claim detection task. Compared to the previous state-of-the-art IMOJIE and Gen2OIE and our baseline method HClaimE, HFE achieves the highest performance with 66% <i>F</i>1 score. This suggests that HFE is a promising framework for various natural language processing (NLP) downstream tasks, such as knowledge graph construction, fact-checking, and question-answering systems. Our dataset and code are publicly available at: <a href="https://github.com/FizaGulzarHussain/CNHeadlines-Corpus">https://github.com/FizaGulzarHussain/CNHeadlines-Corpus</a></p>