Abstract

This paper examines how leading developed jurisdictions protect artificial intelligence related inventions within existing patent law frameworks. It traces the rapid growth and concentration of AI patenting activity and situates it against the TRIPS Agreement obligation to provide patents in all fields of technology, while noting TRIPS silence on the meaning of invention and inventor. The study maps emerging categories of AI related inventions, with particular attention to AI generated inventions, and analyses how core patentability criteria of novelty, inventive step and industrial applicability are applied to data driven and algorithmic technologies. It then conducts a comparative analysis of the United States, the European Patent Office and the United Kingdom on three doctrinal pressure points: inventorship and the requirement of a human inventor, subject matter eligibility for AI algorithms and computer implemented inventions, and sufficiency of disclosure in the context of opaque, black box models and training data. The findings highlight convergence on a human centred conception of inventorship, but divergence on eligibility standards and disclosure expectations, producing legal uncertainty for cross border innovators. The paper concludes by suggesting that clearer guidance on AI specific disclosure and technical contribution is essential to preserve the patent bargain and support balanced AI innovation.

Keywords