PODCAST: Key AI Insights from September Issue of EPI Journal

AI takeaways - EPI - September 2025

16:42

Is AI Rewriting the Definition of Invention?

The rapid adoption of specialized AI co-pilots is forcing the European patent world to confront philosophical questions about obviousness and creativity, fundamentally challenging the definition of the "skilled person". This podcast* episode dives into the current reality of AI use in patent practice, drawing heavily on September's journal from the European Patent Institute.

*podcast generated using the AI tool NotebookLM

Cover of September's EPI journal. Click here to read it integrally.

Quick overview:

AI's Double-Edged Sword:

Patent attorneys are leveraging sophisticated, dedicated AI co-pilots, costing up to €800 per month per attorney, to overhaul drafting workflows and achieve significant efficiency. The tools are often rated highly for generating descriptive text (sometimes up to 80% usable in the context of drafting patent description), but they present major liabilities. A crucial limitation is AI's struggle with drafting claims, frequently including too many non-essential features and accidentally making the scope of protection too narrow.

Ethical concerns are paramount: Due to significant data security and confidentiality risks, attorneys retain full professional responsibility for all output. Furthermore, firms must now seek explicit and specific informed client consent identifying exactly where AI tools will be used, rather than relying on vague blanket clauses.

The Standard Under Threat:

The biggest conceptual conflict revolves around the "person having ordinary skill in the art" (PHOSITA) the legal yardstick used to determine if an invention is legally obvious. The debate asks whether this fictional person must now be assumed to routinely use Large Language Models (LLMs). If the answer is yes (the "augmentation view"), the bar for patentability would be raised dramatically. Currently, the European Patent Office (EPO) is adopting a cautious, traditional view, but the uncertainty remains.

This collision of rapidly advancing, non-deterministic AI technology and static legal principles suggests a potentially radical shift in how inventiveness is judged. The episode concludes by positing a crucial transition for the legal framework:

We could move from assessing human creativity to benchmarking machine capability.

The ultimate open question facing the patent world could soon transform from asking: Could a skilled human have done this without inventive step at priority date? to asking: what were the capabilities of the standard AI systems available at the effective date?

Did you like this podcast and article? If you want to navigate the innovation landscape with clarity, let’s connect here!💡