Insights

A case brief circulated this week by the EUIPO’s anti-infringement centre suggests that the EU’s enforcement playbook is shifting earlier in the chain. According to the account, infringers were no longer relying on one fake sign used repeatedly. Instead, they used generative AI to produce thousands of text-and-image trade mark variants at speed, trying to slip past keyword-based filters before distributing goods across Europe through fragmented cross-border small parcels. The policy response was just as notable: EUIPO said it would flag unusually frequent trade mark variant filings with no credible business logic and work with Member State customs on dynamic interception based on underlying image features.

What makes this development worth watching is not simply that another counterfeit network was disrupted. It is that the governance logic is moving from spotting a bad result to recognizing a bad method. For brands, platforms and cross-border sellers, risk is no longer concentrated only in one listing, one application or one shipment. It is spreading across the full sequence of filing, listing, promotion, dispatch, parcel splitting and customs clearance. Tactics that once bought time by changing one letter, one image or one package profile are starting to lose that advantage.

A newly disclosed ruling from the UPC’s Munich Local Division puts a sharper spotlight on a familiar but often under-examined enforcement tactic: sending broad patent warning letters first and relying on commercial pressure before technical substantiation fully catches up. The message emerging from this development is not that warning letters are improper as such, but that the Court may look far more critically at campaigns built on expansive claim language without a concrete infringement record behind them.

The more interesting shift is procedural. The defendants were able to push a form of reverse interim relief into practical view, signalling that the UPC may be prepared to police not only patent rights, but also the way those rights are asserted in the market. That matters well beyond one dispute. It affects NPE strategies, licensing campaigns, supplier communications and the risk calculus for smaller manufacturers operating across Europe.

Discussion around deferred publication of UK designs has moved faster than the official text. The headline figure catching everyone’s attention is 30 months. Yet the verified position is narrower: the current UK regime still allows deferment for up to 12 months, and the 2025 government consultation compared 12, 18 and 30-month models while signalling a preference for 18 months rather than an already launched 30-month pilot.

Another point is no longer hypothetical. Since 1 April 2026, the DF2C fee to register a previously deferred design has been set at £50. For fast fashion and consumer electronics businesses, the immediate lesson is not simply about how long secrecy may last. It is about when to file, when to keep variants confidential, when to commit to a UK or wider European launch, and how to stop a design portfolio from turning into dead stock on the balance sheet.

WIPO’s Madrid Registry says its Automated Correction System for irregularities, or MACS, has now gone live after technical integration with core European offices. Common formal defects, including loose goods and services wording and format conflicts, are expected to be identified earlier and blocked in real time. On one level, this is a speed story: if fewer routine defects travel downstream, average registration timelines may improve materially.

But that is not the part applicants should focus on first. Once automated checks move to the front end, much less room remains for broad drafting, soft-edged wording and the familiar habit of filing first and cleaning things up later. What used to be managed through later irregularity notices is starting to be controlled at the point of submission.

China’s revised Patent Examination Guidelines, effective from 1 January 2026, move legality review much closer to the centre of AI and big-data patent examination. CNIPA did two important things at once. It added an explicit Article 5(1) review standard for AI and big-data applications, and it also revised the examination baseline so that, where necessary, examiners may review the specification itself rather than looking only at the claims. That is a meaningful shift. Applications involving data collection, label management, rule setting or recommendation decisions are no longer judged only on whether the technical effect sounds persuasive. The file may also be read for obvious legal or ethical fault lines.

The official examples make the point in concrete terms. One concerns a facial-recognition marketing system that did not show lawful and compliant data acquisition. Another concerns an autonomous-driving emergency model trained to differentiate between people by age and sex. For applicants filing inventions around large-model training, corpus cleaning, data-labelling pipelines, alignment methods or vertical-model deployment, the signal is plain enough: training data and data-processing pathways are no longer a background black box the patent file can safely ignore. CNIPA has not published a standalone checklist of acceptable training-data provenance, but the distance between patent entitlement and data governance has clearly narrowed.

IP Australia’s current trade mark manual makes the point more plainly than before: a specification that is unrealistically broad can put intended use in issue at the examination stage. Filing an application will still usually be taken as indicating use or intended use, but that presumption no longer does all the work once the claimed goods or services are commercially implausible on their face.

That matters for applicants who have relied on wide, placeholder-style filings to reserve room first and define the business later. In the digital economy, the familiar combination of classes 9, 35 and 41 can bundle software, platform operations, advertising, training and content services into one application. Where the commercial story behind that bundle is thin, examiners now have clearer ground to demand a declaration of intended use or to press for a narrower specification at the front end.