Insights

Brazil’s National Institute of Industrial Property (INPI) announced on 24 July that DIRMA will hold a public stakeholder meeting on 19 August 2026, with in-person and online participation. One agenda item is the presentation and discussion of revised registrability criteria for position marks, following the close of the related public consultation on 24 July. The direction disclosed by INPI is to assess the sign, its clearly defined placement and the relevant public’s perception of that combination, rather than treating the position in isolation. The final approach should still be checked against the meeting materials and any subsequent formal text.

Acquired distinctiveness is not a mechanism that begins only on 19 August: INPI has already completed consultation work and adjusted its trademark examination framework through measures adopted in 2025. Read together, the two developments point toward heavier reliance on market perception and evidence of use in Brazilian trademark examination. Applicants for non-traditional marks should start organising consistent depictions of the claimed position, duration and geographic scope of use, advertising expenditure, sales materials, surveys and media references. A coherent evidentiary record tied to the relevant goods or services may matter more than arguments based solely on visual originality.

On 1 August 2026, the European Union Intellectual Property Office (EUIPO) issued its first-month examination feedback following the 1 July rollout of the revised EU design framework. With MP4, OBJ, STL and other dynamic or three-dimensional digital formats now accepted, early filings have exposed a practical weakness: visual disclaimers created through blurring, colour masking or similar techniques often fail to remain consistent across animation frames or viewing angles, particularly in files generated by AI tools or exported directly from modelling software.

EUIPO’s message is straightforward. A disclaimer cannot work only in the principal view. If a masked area breaks, shifts or conflicts with the claimed design in any frame or perspective, the examiner may conclude that the scope of protection cannot be determined. The wider range of accepted formats therefore raises, rather than removes, the need for careful pre-filing review. Applicants should test every frame and angle, verify that excluded matter remains visually stable, and retain version records from generation, modelling and export. AI can accelerate production, but it also makes small inconsistencies easier to overlook.

At a cross-border patent revocation hearing that concluded on 31 July 2026, the Unified Patent Court appeared sceptical that vast quantities of material generated by large language models and predictive systems could, by themselves, serve as decisive prior art against novelty. According to the hearing indications described, the court focused on whether the alleged disclosures were genuinely available before the priority date, supported by real technical work, and sufficiently enabling for a skilled person to put the claimed subject matter into practice. Purely theoretical combinations produced through algorithmic enumeration may therefore carry limited weight where they lack a credible route to implementation or reproducible validation.

The hearing stance should not be read as a categorical rule that AI-generated material can never qualify as prior art; the court’s formal position will depend on the written decision and the facts of the case. The practical message is narrower but important: a revocation claimant will need more than volume, plausibility or textual similarity. It must establish public availability, authenticity and technical enablement at the relevant date. Patent owners facing this type of attack should test the evidence on those points rather than treating the generated output as a conventional disclosure. AI can expand the search universe quickly, but it does not remove the legal and technical conditions for a novelty-destroying reference.

Editor’s note (24 September 2026): as of publication we have not located an official announcement of the measures described below; this article is based on industry briefings and is subject to official confirmation.

On 29 July 2026, the China National Intellectual Property Administration (CNIPA) issued supplementary examination guidance for large language models and generative AI inventions, formally bringing training-data compliance into the patent review process. Where an AI model has been pre-trained or fine-tuned on copyright-protected datasets, applicants are expected to include a declaration addressing the legality of the data source. The guidance also introduces an initial “data source whitelist” covering several trusted open-source datasets; applications that clearly rely on listed datasets may avoid an additional copyright review and qualify for a priority examination route.

The practical effect is to move data provenance from a back-office governance issue into the drafting and filing strategy itself. The whitelist could reduce friction, but its value will depend on scope, update frequency and day-to-day examination practice. Applicants should therefore preserve licence terms, dataset versions, acquisition records and internal use logs before filing, rather than trying to reconstruct the evidence during prosecution.

Indonesia’s Directorate General of Intellectual Property (DGIP) moved to the fee and procedural framework under Government Regulation No. 30/2026 on 1 August 2026, alongside an updated electronic filing portal. The change affects official fees and selected examination timelines across patent and trademark procedures. New filings, renewals, recordals and other transactions submitted from the cutover date are charged under the revised schedule, with increases of close to 50% for some items.

The transition guidance gives a narrow reprieve to billing codes generated on or before 31 July but not yet paid: the former fee may still apply if payment is completed before the code expires. Once a code lapses, a new code will generally need to be issued under the revised rates. Applicants should therefore review all unpaid codes immediately, confirm expiry dates and revisit budgets for accelerated examination and other patent or trademark steps. The practical issue is not only higher cost, but also tighter control over payment timing and internal filing instructions.

On 31 July 2026, the Saudi Authority for Intellectual Property (SAIP) issued a dedicated report on intellectual property protection for the esports and gaming sector, alongside the Esports World Cup in Riyadh. The report separates a game into distinct protectable assets: source code and underlying scripts may fall under copyright, user interfaces and character models may qualify for industrial design protection, interactive technologies and hardware innovations may support patent filings, while names, event identities and commercial signs remain within the trademark system. For creators, developers and investors, the practical message is clear: protection should be built as a portfolio rather than around a single right.

The report also gives prominent attention to digital piracy, online dissemination and evidentiary challenges, signalling a compliance and enforcement agenda that runs in parallel with sector promotion. The difficult part will not be identifying possible rights, but documenting ownership and preserving evidence across development cycles. Version histories, code repositories, design iterations and licensing records will often determine whether the framework works in practice. Companies active in Saudi Arabia should therefore review development, publishing, platform and tournament agreements before disputes arise.