THE EAPO IS NOT MERELY A REGULATOR FOR PATENT EXAMINATION, BUT ALSO A CENTRE OF DIGITAL EXCELLENCE AND A KEY COMPONENT OF THE INNOVATION ECOSYSTEM.
– Grigory IVLIEV, President, Eurasian Patent Office, Scientific Supervisor at the Federal Institute of Industrial Property, Candidate of Legal Sciences
TODAY, AI HAS BECOME INTEGRAL TO TECHNICAL SOLUTIONS ACROSS A WIDE RANGE OF INDUSTRIES. THIS SETS AN IMPORTANT TASK FOR PATENT OFFICES— TO ENHANCE THE INTELLECTUAL PROPERTY SYSTEM WITHOUT COMPROMISING LEGAL CERTAINTY.
AI AND ITS IMPACT ON INTELLECTUAL PROPERTY
ARTIFICIAL INTELLIGENCE SUBJECT TO PATENTING
Artificial intelligence (AI) has emerged as a key driver of technological progress. Over the past decade, the number of patent applications for AI inventions worldwide has surged nearly twentyfold. The Eurasian Patent Office (EAPO) has seen the very same trend for several years now, with a steady rise in demand for patenting inventions incorporating AI technologies. AI is no longer a narrow, specialised tool—it has evolved into an infrastructural technology applied across medicine, energy, transport, defence, education, finance, and beyond. It has become integral to complex technical solutions across industries.
In this regard, patent offices face a critical challenge—how to ensure the development of the intellectual property system without compromising legal certainty.
The fundamental principles of patent law remain steadfast. Rather than revisiting the core criteria for patentability, the world is refining and adapting examination practices to ensure they reflect modern technological reality while upholding robust legal certainty.
Leading global patent offices are developing methods for examining technical solutions developed with AI assistance. In many jurisdictions, the key to patentability lies in demonstrating that the AI system addresses a technical problem.
The EAPO is advancing the in-depth research of this far reaching matter. In 2025, in partnership with the Federal Institute of Industrial Property, it conducted a targeted research project “Investigating Issues Related to the Sufficiency of Disclosure in AI-Based Patent and Utility Model Applications.” To illustrate let us examine EAPO’s approach to assessing the patentability of AI inventions.
Sberbank has been granted Eurasian Patent EA 042889 for an invention designed for the automated monitoring of land, forest, and water resources, including in the context of post-disaster recovery. The technical solution relies on analysing data from a satellite constellation using deep learning methods implemented via artificial neural networks. It enables the automatic recognition of agricultural field contours in satellite imagery, going beyond mere identification of ploughlands to accurately divide them into individual fields, even when they visually merge.
In this case, the patentability of the solution rests on a set of essential technical features in the patent claims—a two-stage neural network algorithm for processing spatial data (agricultural land). Unlike existing solutions, which identify only one type of agricultural land, the claimed method delivers precise boundaries of individual fields as discrete objects. Thus, the solution involves the processing of physical data (satellite imagery) designed to achieve a specific practical outcome—the accurate identification of objects on the Earth’s surface.
The core challenge in patenting AI lies in the fact that patent law lacks a clear understanding or definition of the term “artificial intelligence.” Within patent frameworks, AI models are often sought to be protected as inventions in the form of mathematical models, yet mathematical models are not inventions. AI models are also described as devices mimicking the human brain, yet they fall far short of replicating its functionality.
A key criterion for AI-based patentability is technical result. However, the boundary between technical and non-technical outcomes remains blurred.
Patent offices are compelled to address these issues primarily through case law. While justified, this approach risks leaving the regulatory framework trailing behind technological advancements. Such a method may, to some extent, act as an “anchor” for technical progress.
This research project sought to address the question of sufficiency in disclosing AI-based inventions—namely, what must be included in the description to enable its practical implementation. The latter is understood as achieving the declared technical results.
The research yielded concrete proposals to refine the regulatory framework governing the examination of AI-based inventions within the Eurasian patent system. Among these is the recommendation for invention’s description to include an exhaustive set of initial data and requirements that underpin the structure and operational logic of the trained algorithm.
Moreover, established patent law approaches to defining entitled subjects remain unchanged. AI simply does not fit within the framework.
In this context, the recent DABUS (Device for the Autonomous Bootstrapping of Unified Sentience) case is particularly noteworthy. Developed by inventor Stephen Thaler, DABUS is an AI system that operates autonomously, without executing tasks on instruction.
In 2021, the European Patent Office refused to recognise AI as a legal subject, its primary argument being that AI is not a natural person.
In 2023, the UK Supreme Court dismissed an appeal by Stephen Thaler, stressing that, under the law, only a natural person can be recognised as “inventor.” In June 2025, the Swiss Federal Administrative Court ruled that an AI system operating on the DABUS principle cannot be named inventor in a patent application. The court held that if a natural person files a patent application for an invention created by AI, it is that individual who may be recognised as the inventor.
EURASIAN PATENT SYSTEM
The Eurasian patent system stands as one of the world’s largest regional patent frameworks, operating under the Eurasian Patent Convention, signed on September 9, 1994. It provides legal protection for inventions and industrial designs across the vast expanse of the Eurasian continent—encompassing eight countries: Turkmenistan, Belarus, Tajikistan, Russia, Kazakhstan, Azerbaijan, Kyrgyzstan, and Armenia.
For over thirty years, the Eurasian patent system has consistently remained in high demand. Today, it is used by applicants from more than 130 countries worldwide.
The system is built around the one-stop shop principle: all stages of examination and the granting of a single patent are handled by the EAPO. Applicants submit a single form in one language (Russian) and pay a single set of fees in one currency. EAPO’s global standard of quality in examining Eurasian applications is certified by ISO 9001:2015.
AI AS INTELLECTUAL SUPPORT INFRASTRUCTURE FOR PATENT OFFICE
Today, the EAPO views AI not only as a subject of patenting but also as a critical infrastructure, enabling the office to operate more swiftly, precisely, and efficiently.
Crucially, AI does not replace or override the decisions of human examiners: intellectual property examination has always—and will always—be the responsibility of humans. For EAPO, AI is an intellectual support infrastructure, helping examiners cope with ever-growing volumes of data, reduce the burden of routine workloads, and focus on substantive examination of patent applications.
AI also paves the way for streamlining patent timelines through smarter workflow and a more efficient use of specialist expertise—without compromising the quality or legal reliability of the examination.
The integration of AI within the EAPO is carried out in a systematic and tightly governed manner. In 2025, the Artificial Intelligence Technology Implementation Centre was established to consolidate AI developments, examiner approaches, and practical applications in patent procedures. This format enables the establishment of uniform standards and ensures that implemented solutions align with legal and methodological requirements.
As part of this initiative, a prototype AI assistant for examiners has recently been introduced. This intelligent tool is designed to support every phase of an examiner’s analytical workflow—from document search and analysis to drawing structured conclusions and drafting decision proposals. By automating routine tasks, taken to an entirely new level, this system allows examiners to focus fully on the substance of technical and legal analysis.
Other tasks for which the EAPO is developing and deploying AI solutions include automated translation of patent documentation, patent classification, searching for similar patent documents and industrial design images, distributing workload among examination units, and providing advice to users on Eurasian patent regulations.
AI will soon address tasks in patent analytics, examination quality monitoring, screening submitted materials, and structuring applications upon submission.
As experience shows, AI is penetrating virtually every area of office operations, transforming its work from within at an unprecedented scale. In effect, a uniform AI infrastructure is taking shape, integrating software, hardware, and specialist expertise. For the Eurasian region, data exchange is particularly vital, hence the advocacy for full scale integration. Interlinking systems helps avoid duplication of work and reduces costs for each office. Thus, for instance, the Eurasian Accelerated Patent Prosecution Programme (EA PPH) marks a step forward for the entire region.
These new capabilities are not only reshaping internal processes but are also gradually transforming intellectual property systems as a whole.
For instance, AI-powered search for similar images of industrial designs makes it easier for offices around the world to transition from a registration based system to a robust examination framework, significantly enhancing the reliability of such protection.
In the age of AI, the very role of the patent office is undergoing change.
The modern patent office, such as the EAPO, is no longer merely a regulator or centre for patent examination. It is also a vital hub for aggregating and processing vast volumes of scientific and technical data, a centre of digital excellence, and a key component of the innovation ecosystem.
EAPO’s task is to be an active participant in this technological transformation, developing a systematic approach to integrating AI and ensuring the sustainable development of the intellectual property system amid rapid technological change.
AI is a rapidly evolving field. For EAPO, it is a means of generating new solutions applied across all scientific and technical domains. It is a tool, a method, a technical solution. The rights to inventions, utility models, industrial designs, and other creations developed with AI will belong to a legal or natural person—a critical consideration not only in terms of rights but also regarding accountability. AI itself is a tool, that is its essence, and it is through this lens that it should be understood and used—as one of the most advanced tools of the 21st century.
ADVERTISEMENT. Legal entity Eurasian Patent Organization, INN 9909057949, ERID 2SDnjcd5LMP