Warning on AI Misuse: Japanese IP Court's Rejection of Statutory XML Gazette Raises Doubts Over Judgment Basis

Overview

The Intellectual Property High Court of Japan has been revealed to have ruled not on the officially recognized statutory gazette format of the Patent Office, but instead on a "standalone PDF gazette" provided by a third-party platform, in litigation involving Patent No. 7105571 (plaintiff PACRAFT Co., Ltd.). This incident has triggered dual concerns within academia and industry regarding the rigor of judicial proceedings and the risks of abuse of AI-generated content.

Gazettes in Controversy in Judicial Practice

The Japan Patent Office (JPO) explicitly stated that the "statutory gazette" issued after January 12, 2022, refers exclusively to XML format files provided via the gazette issuance website. However, in the judgment for Case No. "Reiwa 5 (Gyo-KE) No. 10092," the "Exhibit 1" cited by the court was not the official XML gazette, but a PDF version submitted by the plaintiff.

Upon comparison, this PDF matched the "standalone PDF gazette" provided in the J-PlatPat database operated by INPIT exactly. The judge cited this document on page 2 of the judgment as the basis for determining technical features, yet failed to display the specific content of "Exhibit 1" in the public version, rendering it impossible for the public to verify the basis of the trial.

The "standalone PDF gazette" is merely reference material and does not possess the legal status of a statutory gazette.

Hidden Dangers Arising from the Lack of Data Source Standardization

A deeper issue lies in the chaos of data sources. The JPO has emphasized that the XML format is the only statutory standard, while multiple private service providers, including INPIT, have launched their own formats of PDF gazettes. Comparisons show that even for the same patent, the PDFs provided by different vendors (such as Hitachi System's SRPARTNER and J-PlatPat) exhibit significant differences in page count, layout formatting, and even claim text.

This situation of non-unified standards means that judicial determinations may be built upon non-authoritative, non-original tertiary sources. If parties rely on AI tools to automatically scrape and analyze these "standalone PDFs," legal arguments are highly prone to errors due to data bias.

Industry Reflection: Beware of AI Dependency

This case has exposed the blind spots of professionals when utilizing digital tools. Many practitioners are accustomed to directly downloading ready-made PDF files from search engines or commercial platforms, ignoring the issue of data traceability behind them.

The JPO has called for all legal discussions involving patent validity to return to the original statutory gazette in XML format. Intellectual property practitioners should abandon blind trust in content generated by commercial platforms, strengthen their awareness of verifying original legal documents, and prevent damage to judicial fairness caused by the misuse of tools.