4.14.2026

[International IP Briefing] US, EP

 

US

USITC Institutes Section 337 Investigation into DRAM Memory Semiconductor Companies, Including SK hynix

On March 26, 2026, the U.S. International Trade Commission (ITC) announced the institution of an investigation under Section 337 of the Tariff Act. The investigation is based on allegations that certain NAND1)  and DRAM memory semiconductor technologies from companies including South Korea's SK hynix and Japan's Kioxia2) infringe upon the patent rights of MonolithIC 3D, a U.S.-based patent licensing firm.

 (Background)

Section 337 of the U.S. Tariff Act grants the ITC quasi-judicial investigative powers regarding unfair practices in import trade, including the infringement of intellectual property rights. Under this provision, the ITC may issue Exclusion Orders and Cease and Desist Orders against goods found to be in violation.

-      The ITC determines the target date for completing the investigation within 45 days of its institution.

-      Unless the Office of the United States Trade Representative (USTR) disapproves the ITC’s decision within 60 days, the decision takes immediate effect as an administrative order.

(Key Details)

The specific details regarding the background and the decision to institute this investigation are as follows:

1. Background of the Investigation

On February 17, 2026, MonolithIC 3D3) filed a complaint with the ITC requesting a Section 337 investigation, asserting that certain NAND and DRAM memory semiconductor technologies imported into and sold in the U.S. infringed its patents.

Patents at issue:

2. Decision to Institute Investigation

·         Respondents: The ITC has decided to commence a Section 337 investigation targeting South Korea's SK hynix and Japan's Kioxia, among others.

·         Procedure: The ITC will assign an Administrative Law Judge (ALJ) to preside over the evidentiary hearing and related proceedings. Following these procedures, the ALJ will issue an Initial Determination on whether there is a violation of Section 337, which the Commission will then review for a final decision.

 

-----------------------------------------------------------------------------

1) NAND is a type of flash memory that has the characteristic of retaining stored data without a power supply.

2) Kioxia Holdings Corporation, Kioxia Corporation, Kioxia America, Kioxia Engineering Corporation, Kioxia Iwate Corporation, Kioxia Systems Co.

3) MonolithIC 3D additionally amended and supplemented the contents of the complaint on March 16, 2026.


< Source of this post >

https://www.kiip.re.kr/board/trend/view.do?bd_gb=trend&bd_cd=1&bd_item=0&po_item_gb=US&po_no=24340

 

EP

European Patent Office Integrates AI-Powered OCR Solutions into Operational Systems

In March 2026, the European Patent Office (EPO) announced that it had co-developed and integrated an AI-powered Optical Character Recognition (OCR)1)  solution into its operational systems in collaboration with the European AI firm, Mistral AI.

 (Background and Overview)

The EPO has long utilized OCR technology to extract text from patent documents submitted in paper or PDF formats, transforming them into searchable and analyzable data. However, modern patent documents often contain complex information—including multilingual text, mathematical formulas, chemical structures, images, and tables—that traditional OCR technology struggles to capture accurately.

These limitations can lead to recognition errors or data omissions during the conversion into structured data, which in turn hampers the efficiency of patent examinations, prior art searches, and overall analysis quality.

 As part of its digital transformation under the AI Policy2)  and Strategic Plan 20283) , the EPO initiated this partnership with Mistral AI to:

Strengthen European Digital Sovereignty by leveraging AI technologies developed by European enterprises.

②Ensure Data Security, guaranteeing that sensitive patent information is processed according to European legal and ethical standards.

(Key Highlights)

Following a three-month Proof of Concept (PoC)4)  tailored to the specific characteristics of patent documents, the EPO and Mistral AI developed a specialized AI-powered OCR solution. Its integration into the EPO’s workflow is expected to yield the following benefits:

Enhanced Accuracy in Data Extraction: The solution significantly improves the recognition of complex elements like formulas, chemical structures, and tables, ensuring precise extraction of technical information used in the patent granting process.

Strengthened Foundation for Patent Analysis: By effectively converting unstructured information into structured data, the EPO establishes a robust data foundation for patent management and technical trend analysis.

Improved Efficiency in Examination and Prior Art Searches: Structuring previously difficult-to-analyze documents allows EPO patent examiners to conduct prior art searches and technical evaluations more efficiently.

Future Plans

The EPO plans to continuously refine the AI-powered OCR solution and expand its scope of application. Moving forward, the Office will explore the introduction of additional AI technologies to further optimize operations across the entire patent system.

--------------------------------------------

1) Optical Character Recognition (OCR): A technology that automatically recognizes characters in scanned documents or images and converts them into digital text data, enabling document search, analysis, and data utilization.

2) AI Policy: Established in February 2025, this policy creates a framework for the responsible use and governance of AI. It guides the EPO's implementation of AI in tasks such as patent searching and document analysis. (https://link.epo.org/web/about-us/transparency-portal/en-epo-ai-policy.pdf)

3) Strategic Plan 2028: For the original text of this plan, please refer to the following https://link.epo.org/web/about-us/office/en-epo-strategic-plan-2028.pdf

4) Proof of Concept (PoC): A validation phase where a technology or solution is implemented on a limited scale to verify that its functions and performance meet the requirements of a real-world environment.1.

 

< Source of this post >

https://www.kiip.re.kr/board/trend/view.do?bd_gb=trend&bd_cd=1&bd_item=0&po_item_gb=EU&po_no=24300










No comments:

Post a Comment