In Practice Topic 1, the theme was “Anticipating and addressing rejections of claims to AI inventions in different jurisdictions”.
Well-guided by the moderator Roberto Pistolesi, President of FICPI and Managing Partner of Dragotti & Associati (IT), patent attorneys Sharon Crane of Haynes & Boone (US), Andreas Oser of Prüfer & Partner (DE), and JiHoon Joseph Hong of Invensync (KR) reported on IPO practices and explained specific requirements and amendments how to overcome rejections that AI-related inventions usually face across jurisdictions, referring here to the US, Europe and Korea, respectively. Same patent families that were actually prosecuted before the corresponding Patent Offices were selected as study examples, allowing to speakers to better compare and understand the common, but also the distinct approaches applied in the various jurisdictions.
Interestingly, while naturally there are also common grounds, quite distinctive practices could be identified when it comes to specific individual patenting requirements.
Before the USPTO, the “101” patent eligibility requirement, with its recently sharpened step-wise tests, poses a critical hurdle to overcome, Sharon explained. Practitioners are advised to make careful own considerations on these eligibility tests, to thereby overcome – and better anticipate before filing – such problems during prosecution.
Addressing the corresponding cases before the EPO, Andreas touched upon the EPO’s updated examination guidelines to clarify the minimum requirements on the technical character of AI related inventions. According to the case examples discussed by Andreas and then also by Joseph, the EP and Korean practices seem quite similar, in that a disclosure of the case-specific correlation between input data and output data, how the implemented AI model is trained, and how the correlation or the output serves to solve a specific technical problem appear as most relevant issues.
On the other hand, examiners of the USPTO tend to transfer considerations, which otherwise rather belong to substantive assessments, into the eligibility questions, in particular whether there is teaching beyond a mere mathematical concept and providing “significantly more than the judicial exceptions”. In the related situations before the EPO and the KIPO, Applicants of AI-related inventions are advised to disclose, in sufficient manner, how the AI system is trained and how this addresses the intended purpose and solves the underlying problem, to thereby overcome corresponding objections by suitable amendment – so the original specification does provide sufficient disclosure thereon.
The discussion with the audience then addressed further relevant questions of interest, such as how to overcome AI-related objections in addition to amendments (then especially explaining the admissibility of post-filed evidence) and the challenges of enforcing AI-related patent claims.
Thus, after the 90 minutes of intense discussions among all, the perspectives on AI-related inventions from across various jurisdictions became clearer.
FICPI's view and involvement
FICPI uniquely combines education and advocacy on topics around patents and trade marks, with a focus on developing the professional excellence of its individual members. FICPI Fora, Congresses, committees and meetings are opportunities to gather insights from the international IP attorney community on any issue, whether it be practice-related or topics of patent and trade mark law.
Next steps
If you’re interested in finding out more about the work of FICPI’s Committees or becoming involved, visit: https://ficpi.org/organisation/committees/
Join us at our 2025 World Congress in Naples, Italy from 12-17 October 2025.
