On April 30, 2025, the China National Intellectual Property Administration (CNIPA) released a draft amendment of the Patent Examination Guidelines (for public comment), proposing further revisions to...

It is fair to say that no technology in recent times has created so much excitement and controversy as generative AI, and in particular Large Language Models (LLMs). Generative AI companies and their backers have their sights on generating billions, if not trillions of dollars, directly or indirectly from their models. Models which, in the vast majority of cases, are trained on original works without permission from, or payment to, their authors. The works used for training are primarily obtained by scraping the open web, a uniquely vast repository of publicly accessible knowledge and information, constantly updated to reflect the latest developments in human knowledge and culture.

On 2 July 2025, the Enlarged Board of Appeal of the European Patent Office (EPO) issued its landmark decision in case G 1/23 (Solar Cell Sealing Materials and Solar Cell Module), a ruling that has...

For deep tech startups, intellectual property decisions often come too late or cost too much too soon. This article explores why IP strategy should be considered from day one, even if formal protection comes later. It outlines how founders can decide what to patent, when trade secrets may be more appropriate, and how detectability, enforceability and cost should shape early choices. With a focus on practicality and commercial alignment, it also looks at how newer, AI‑assisted approaches can help startups take a more cost‑effective and flexible approach to building meaningful IP protection.

Recent US court decisions have, for the first time, tested whether conversations with AI tools can be protected by legal privilege. In United States v. Heppner, a court held that documents generated by a defendant using a public AI platform were not protected by attorney‑client privilege, largely because the AI was not used within a lawyer‑directed relationship and the platform’s terms allowed data retention and disclosure. By contrast, in Warner v. Gilbarco, another court found that AI‑assisted materials created by a self‑represented litigant were protected under the work‑product doctrine, which has a more forgiving standard for waiver.

On the Technically Legal podcast, EIP partner Laurence Brown discusses how Visser AI is redesigning patent workflows, letting the firm offer fixed fees without trading away speed or quality.

This article looks back on a selection of key trends in the (bio)pharma industry in 2025 and looks ahead to how these may continue to develop into 2026.

As AI use accelerates in patent practice, many European patent attorneys remain cautious due to professional AI guidelines from the epi. This article argues those guidelines are not a barrier, but a baseline for responsible use.