Smarter Patent Intelligence: How AI Identifies Potential Infringers with Greater Accuracy

Introduction

The constant inception of new technologies by industries puts patents at risk. Having more than 3.7 million patents means IP attorneys struggle to manually assess whether someone has infringed on their invention. Despite the traditional methods, the challenge and patent avalanche are too great. The current era is only beginning to patent inventions, and the actual value of an invention comes from the fact that the rights of the patentee are protected by their right to use them involuntarily without authorization.

Conventional methods for detecting patent violations have been time-consuming and laborious. To identify potential overlaps, patent attorneys and technical experts examined complex claim language from competing products. But moving markets are changing rapidly and this manual approach is no longer sustainable.

The use of intelligence-based IP strategies is on the rise in recent times. Advancing market surveillance through artificial intelligence (AI) and machine learning (ML) algorithms is being utilized by companies. Stakeholders can identify potential breaches with greater accuracy and save money by utilizing automated tools for infringement detection. I am writing here with an interest to explain how AI-powered patent intelligence works : from the perspective of natural language processing (NLP) models of interpreting patent claims themselves, and also from exploring the relationship between algorithms and technical resources, including whether specialized human-AI hybrid evaluation models are necessary to achieve successful implementation.

Legal Provisions

The Patents Act of 1970 was the catalyst for the patent infringement legislation in India. The patentee is authorized by Section - 48 of the Act to prevent others from creating, using, and importing the patented product or process without their consent. Patent claims solely serve as evidence of infringement, as they define the invention's scope. The plaintiff in a patent lawsuit must prove that the defendant's product or process is within their rights under Section 104 of the Patents Act, 1970. What are the following provisions? There exists an ongoing process called 'claim construction' or simply - "and we call it... interpretation.".

Legal Analysis.

• AI Models for Claim Interpretation : Claims are the primary legal obstacle in infringement analysis. Patent claims are written in a legal language that's formulaic. Self-limiting claims set the boundaries of protection, whereas dependent claims limit them. Why? Modern AI models, which incorporate NLP and Large Language Models (LLMs), have made this process unique. This is significant. These models employ parsing to divide independent and dependent claims into distinct technical concepts.

• The integration with Product Databases and Public : Technical Resources can only be achieved through the analysis of terminology. The procedure can be explained in greater detail later. Patent claims are broken down into smaller pieces using Artificial Intelligence, which can interpret technical and legal texts. The capabilities and limitations of an inventor are outlined in these sentences. Later on, Artificial Intelligence investigates what' s presently existing in the world. By utilizing databases and public resources like engineering forums and academic journals, Artificial Intelligence can gain an understanding of other companies' products and technologies.

  
  

  Artificial Intelligence uses web scraping algorithms to collect all the information. Similar to computer programs that enter data from websites and gather it automatically, they analyse a wide range of documents, including user manuals and product information sheets. ". They scrutinize policies that dictate the most effective approach for products to integrate. To verify compliance, Artificial Intelligence examines submitted government documents by companies. The group scrutinizes the promotional materials utilized by businesses. These documents frequently describe the properties of the products, which could be similar to those in patent claims. All of the data is evaluated by Artificial Intelligence to understand its capabilities and patents.

An unprecedented level of scale-up of analysis for the AI is facilitated by the integration. The AI will examine the hardware components of competing products that meet the independent claim for a biometric sensor to measure blood oxygenation. Additionally, The use of Pattern Recognition for Infringement LikelihoodAI in machine learning algorithms enables the calculation of an infringement likelihood score, rather than relying on binary determination.

• Human-AI Hybrid Evaluation : Law and fact make patent infringement sensitive, but the capabilities of AI in scanning and parsing make it a sensitive issue. It is possible that AI does not always adhere to this pattern, as an example is the impossibility of understanding the prosecution history of a patent in relation to specific claims.

  
  

  As a result, an AI hybrid evaluation model is necessary. The AI functions as a productive primary label, distinguishing between actionable intelligence and highlighting the precise technical mapping between the claim exemption to the accused product. The Markman case highlights the legal requirements for human expert validation.

 Case Laws.

• Markman v. Westview Instruments, Inc, 517. U.S. 370. : In the US Supreme Court case, patent claims were not interpreted by juries but rather by judges. While it's possible to use AI to refute the notion that a human judge is at fault, this evidence emphasizes its practicality.

Automat Irrigation Pvt.. In Ors v, Aquestia Limited & Anr. was heard in the Delhi High Court in 2026.

Clarification on the legal interpretation of patent claims was provided by Delhi High Court. The reason behind their position was to scrutinize the whole matter and not just a specific part. India's claim charts require AI systems to examine the entire claim rather than just a single segment, as there are other countries with similar conditions.

Plant-e v. Bioo (Unified Patent Court. The Hague Local Division, 2024) : The Doctrine of Equivalents was established by the Unified Patent Court.... A patent was examined for the production of electricity by living plants using microorganisms. No living plants were present in the defendant's product, so it was not against the law. Four-stage review was instituted by the court. The test asks:

Does the variant provide a solution to the technical equivalence problem?

Is there a benefit to expanding the patent protection period?

Is the patent reliable enough to be interpreted with reasonable confidence by external sources? Does the opposite notion not have a distinct interpretation?

Despite the defendants' system being breached, the court determined that it was technically identical. Through this case, the Doctrine of Equivalents is being brought across state borders. Why? These criteria have been made mandatory for AI patent tools that help identify European competitors.'

FMC Corporation v. Natco Pharma Limited (Delhi High Court, 2022) : The process patent of Natco for creating a chemical was violated in the lawsuit filed by FMC against the company. Natco used a method. The Doctrine of Equivalents was disputed by FMC.? Why? A court's examination of the patent specification... Discovered that FMC resorted to one method. The court restricted the claim to that approach and found no breach.' AI tools are required to scrutinize the patent specification and history, not the claims, as per the ruling. If a patentee's words restrict the claim to just enough terms, it becomes difficult for them to extend it further.

Legal Implications

The application of Artificial Intelligence to comprehend them has significant implications for businesses and patent holders.' Why? It mostly helps to move away from the practice of storing patents and towards using them for profit. Through market surveillance, patent holders can determine whether their idea is being exploited or misused. Patent holders earn revenue from the use of Artificial Intelligence.? A company can use Artificial Intelligence to examine patents and determine the appropriate actions to take if someone is exploiting their idea. By utilizing AI, companies can establish a communication bridge and reach an agreement with those who are using their idea without permission.

This helps businesses avoid legal trouble. The integration of Patent law with Artificial Intelligence is an important factor. Patent holders can monitor market events through the use of Artificial Intelligence. This can be attributed to Artificial Intelligence's aptitude to scrutinize patents and products. The validity of patents was once checked in the past. With Artificial Intelligence, individuals can always be ahead of the game regarding market updates. The urgency of the situation is making it imperative to act.

Conclusion

The fact that Artificial Intelligence is a tool should not be overlooked. But it does not replace those who know law. The task requires the assistance of legal professionals to ensure proper execution. Individuals who create will increase protection and efficiency of ideas. Patent holders will receive assistance from Artificial Intelligence and patent law. The utilization of Artificial Intelligence will enable patent holders to comprehend market trends and make informed decisions.

Author: Vatsal Pare, in case of any queries please contact/write back to us via email to chhavi@khuranaandkhurana.com or at  Khurana & Khurana, Advocates and IP Attorney.

References

1. Automat Irrigation Pvt. Ltd. v. The High Court of Delhi's Aquestia Ltd. was established in 2026.

2. F. In 225, Cipla Ltd. was sued by Hoffmann-La Roche Ltd." & Anr. as well. DLT 391 (2015).

3. FMC Corp. v. Natco Pharma Ltd was sued in the High Court of Delhi in 2022.

4. Markman v. Westview Instruments, Inc, 517. U.S. 370. Plant-e v. Bioo is situated in The Hague Landing's Unified Patent L. Div. (2024).

5. The Patents Act, No. 39. World Intellectual Property Organization. (2020). WIPO/IP/AI/2/GE/20/1 paper on artificial intelligence and intellectual property.