Insights

IP Australia’s current trade mark practice makes one point much clearer than before: in a three-year non-use dispute, it is not enough for a brand to point to a product page that happened to be visible from Australia. The harder question is whether the page, payment flow, delivery settings and surrounding marketing actually show that Australian consumers were being targeted. For cross-border e-commerce sellers, that is not a minor evidentiary tweak. It changes what kinds of digital records are likely to matter when a registration is challenged.

The point is not that online evidence suddenly became easy. It is that the evidentiary centre of gravity has moved toward verifiable market direction. A screenshot alone will rarely carry the day. A stronger file is one that can connect page captures, currency settings, shipping options, order logs, ad-targeting reports and customer records in a way that shows genuine commercial activity aimed at Australia rather than a generic storefront floating somewhere offshore.

Recent JPO guidance and examination practice mean virtual 3D assets in Japan can no longer be treated as one undifferentiated bucket of “digital content.” Digital fashion items, scene models, tradable spatial decorations and service-linked 3D displays may all live inside virtual environments, but that does not mean they travel through the same legal door. For some assets, copyright may still be the more natural starting point. For others, the more important question is whether the current Design Act framework offers a cleaner and more enforceable route.

The real shift is not academic. It shows up when rights holders ask platforms to act. Japan’s newer platform-liability framework is pushing large operators to formalize contact points, review standards and response systems, but the logic behind provider exemptions has not disappeared. In practice, anyone seeking faster removal of allegedly infringing virtual goods or spatial assets will need more than screenshots and conclusions. The rights basis, the target asset and the evidence trail now matter much more.

The easiest mistake to make here is to describe recent PTAB developments as if the Office had already issued a stand-alone rule aimed specifically at AI-generated prior art. The more accurate picture is narrower and more important. Through a set of concrete procedural moves, the USPTO has started pushing harder on three questions that matter in inter partes review: where the asserted art came from, whether it really qualifies as a patent or printed publication, and how far a petitioner should be expected to explain its search path. The July 31, 2025 memorandum enforcing Rule 104(b)(4) made clear that petitioners may not use applicant admitted prior art, expert testimony, common sense or other forms of general knowledge that are not themselves patents or printed publications to supply a missing claim limitation. Then, on November 17, 2025, the Office introduced an optional Search Disclosure Declaration process that allows petitioners to describe the databases, repositories, filters and general query logic used to locate asserted art.

That is not yet an AI-specific mandatory disclosure code. But it is a strong directional signal. For parties that now rely on AI-assisted search, semantic retrieval and large-model summarization to identify prior art and non-patent literature, the real problem is no longer just whether more references can be found. It is whether search leads can be kept separate from admissible evidence, whether machine-generated synthesis is being mistaken for a printed publication, and whether the resulting record can survive a PTAB challenge. AI has made prior-art hunting faster. PTAB is making the question of what exactly was found much harder to gloss over.

Argentina’s partial non-use cancellation mechanism is no longer just a technical footnote in the statute. By 2026, it is starting to matter as a practical portfolio and clearance tool. For years, brand owners could often take comfort from the idea that some use within a class helped preserve a wider registration footprint. That comfort is thinning as Argentina’s trademark system shifts more responsibility onto rights holders to watch the register and defend the real scope of their filings.

The issue in practice is no longer simply whether a mark has been used at all. It is where, and for which listed goods or services, that use can genuinely be shown. Once a registration has been on the books for at least five years, a party with legitimate interest may ask INPI to cancel it for non-use, and the partial-cancellation angle turns that proceeding into a sharper instrument. Owners whose evidence really supports only a narrow commercial core may find the unused tail of the specification under pressure.

As of now, the U.S. Copyright Office is still publicly pointing to the May 2025 pre-publication version of Copyright and Artificial Intelligence, Part 3: Generative AI Training, not to a formally effective final AI-training rule. Even so, the policy debate has become much more operational over the past year. If commercial model training continues to rely on platform datasets, open-web scraping and third-party licensing, questions about data provenance, authorization chains and the execution of rightsholder opt-outs are no longer abstract talking points. They are turning into concrete compliance issues that companies may be asked to explain.

For short-video platforms, social media services and other UGC-heavy businesses, the real pressure is not just whether a front-end “no AI training” option exists. It is whether that instruction can actually travel through scraping controls, API outputs, data-sharing arrangements and downstream development workflows. The Copyright Office has not publicly issued a finalized mandatory technical rule, but the direction is already clear: a platform that enables AI training access while struggling to show how opt-outs, license scope and protected-content filtering are implemented will have a harder time presenting itself as a neutral conduit in future copyright disputes.

Brazil’s INPI is starting to make its industrial design acceleration policy more concrete. Under the latest arrangement, industrial design applications designating Brazil through the Hague System may file a free fast-track request online where the product design is clearly tied to environmental efficiency or accessibility-oriented assistive use. For qualifying cases, the target is a substantive review and decision within 30 days. For applicants already treating Brazil as a serious design market rather than a distant filing add-on, that is not a minor procedural perk. It can materially affect launch sequencing, disclosure timing and enforcement readiness.

The broader signal matters just as much as the speed promise. Brazil is not using the Hague route only as a cheaper international filing corridor. It is also beginning to connect design administration with policy goals around green transition and inclusive innovation. Once priority treatment is tied to those themes, applicants can no longer assume that a visually distinctive design is enough on its own. Product framing, use context and the way supporting materials are assembled all become more strategic.