The conversation around AI in the legal profession has been dominated by fear: “hallucinations”, disciplinary cases, and talk of “robots replacing lawyers”.

In IP practice, those fears are amplified by strict confidentiality requirements and the central importance of novelty and inventorship. 

Many attorneys have responded by avoiding AI entirely – even as the technology has rapidly improved and become an everyday productivity tool in other professions.

This session (suggested by Group 6 of FICPI's Study & Work Committee (CET)) will try to cut through the hype and the horror stories to give IP practitioners a clear, candid understanding of what modern large language models (LLMs) are and are not, how they actually work, and how data is handled when you use them. 

The session will explore, in practical terms, whether using AI tools to help draft patent specifications or trade mark advice really risks destroying novelty or breaching confidentiality – and how to design workflows that stay well within ethical and regulatory guardrails.