From Korean “Technical Ideas” to U.S. “Practical Applications”: § 101 Strategy for Software and AI Patent Applications

Executive Summary: The article explains why Korean software and AI patent applications that appear eligible under Korean law may still face significant obstacles under U.S. patent-eligibility doctrine. Korean law focuses on whether the claimed invention is a “technical idea utilizing laws of nature” and, for software-related inventions, whether software-based information processing is concretely realized using hardware. U.S. law, by contrast, asks whether the claim falls within a statutory category and avoids—or sufficiently applies—judicial exceptions such as abstract ideas, laws of nature, and natural phenomena. The article uses Korean Supreme Court and Patent Court decisions alongside U.S. Supreme Court and Federal Circuit authority to show that merely reciting a server, processor, database, AI model, or computer-readable medium may be enough to support eligibility in some Korean contexts but may not satisfy U.S. § 101 if the claim is characterized as data collection, analysis, business decision-making, mental activity, or generic computer implementation. Its practical takeaway is that Korean applicants should not treat U.S. filing as a translation exercise: the U.S. application should identify a concrete technological problem, disclose the technical mechanism that solves it, and recite in the claims—not only the specification—the specific technological improvement or practical application that prevents the invention from being treated as an abstract idea

I. Introduction

For Korean patent attorneys, U.S. patent eligibility under 35 U.S.C. § 101 often appears familiar at first glance. Both Korean and U.S. law contain a threshold inquiry into whether claimed subject matter is the kind of thing that may be protected by patent law. Both systems also show concern for claims that merely recite mental activity, mathematical logic, business rules, or information presentation. Yet the similarity can be misleading. The Korean eligibility inquiry and the U.S. § 101 inquiry are not doctrinal equivalents.

Korean law begins with the Patent Act’s definition of an “invention” as a “highly advanced creation of technical ideas utilizing the law of nature,” together with the requirement that the invention be industrially applicable. Patent Act arts. 2(1), 29(1) (S. Kor.); Requirements for Patentability issued by KIPO, ch. 1, §§ 1, 4 (Jan. 2010). In software and business-method cases, Korean courts have developed a practical inquiry focused on whether software-based information processing is concretely realized using hardware. Daebeobwon [Supreme Court], 2007Hu265, Dec. 24, 2008 (S. Kor.).

U.S. law begins differently. Section 101 provides that a person who invents or discovers a new and useful “process, machine, manufacture, or composition of matter,” or an improvement thereof, may obtain a patent, subject to the other requirements of Title 35. 35 U.S.C. § 101. But in modern software, AI, fintech, business-method, and diagnostic cases, the central U.S. issue is not merely whether the claim is drafted as a process, system, apparatus, or computer-readable medium. The central issue is whether the claim is directed to a judicial exception—especially an abstract idea—and, if so, whether the claim is eligible under Alice/Mayo because it is a patent-eligible application rather than a claim to the exception itself. Alice Corp. Pty. Ltd. v. CLS Bank Int’l, 573 U.S. 208, 217–18 (2014); Mayo Collaborative Servs. v. Prometheus Labs., Inc., 566 U.S. 66, 72–73 (2012).

A drafting warning should be stated at the outset: in U.S. practice, the claims must carry the eligibility story. The specification is important because courts read the claims in light of the specification, but a technical narrative in the specification does not save an abstract independent claim. The Federal Circuit has repeatedly stated that eligibility depends on what is claimed, not everything disclosed in the specification, and that the claim itself must identify how the functional result is achieved at some level of concreteness. See AI Visualize, Inc. v. Nuance Commc’ns, Inc., 97 F.4th 1371, 1378–81 (Fed. Cir. 2024); Trs. of Columbia Univ. in the City of N.Y. v. Gen Digital Inc., No. 2024-1243, slip op. at 8–10 (Fed. Cir. Mar. 11, 2026).

This difference matters for Korean-origin applications. A claim that appears eligible in Korea because it concretely recites a server, processor, database, terminal, or computer-readable medium may still be vulnerable in the United States if a U.S. examiner or court characterizes it as collecting data, analyzing data, applying a business rule, making a mental judgment, calculating a score, or displaying a result on generic computer hardware. The practical drafting task for Korean applicants is therefore not merely to translate the Korean claim into English. It is to translate the Korean “technical idea utilizing laws of nature” into a U.S. § 101 narrative: a concrete technological solution to a technological problem, disclosed in the specification and recited in the claims.

II. The Korean Baseline: “Technical Ideas Utilizing Laws of Nature”

Korean patent eligibility starts from the statutory definition of an invention. Article 2 of the Korean Patent Act defines an invention as a highly advanced creation of technical ideas utilizing laws of nature, while Article 29(1) requires that the invention be industrially applicable. Patent Act arts. 2(1), 29(1) (S. Kor.). KIPO’s patentability guidance explains that Article 29(1) requires both a statutory invention and industrial applicability, and that the “highly advanced” language is generally disregarded when determining whether claimed subject matter is statutory. Requirements for Patentability issued by KIPO, ch. 1, §§ 1, 3–4 (Jan. 2010).

Korean courts treat this statutory definition as a real threshold requirement. In the perpetual-motion-type case, the Supreme Court affirmed rejection of an application directed to a water-powered energy system because the claimed system would have recirculated water without external energy in violation of the law of conservation of energy. Daebeobwon [Supreme Court], 98Hu744, Sept. 4, 1998 (S. Kor.). The case illustrates that Korea’s “law of nature” requirement is not merely decorative language. If the claim contradicts natural law or does not use natural law at all, it may fail at the threshold.

The Korean doctrine becomes more nuanced in software and business-method cases. KIPO’s guidance states that economic laws, mathematical methods, logic, arbitrary arrangements, and mental activities are generally non-statutory when claimed as such. Requirements for Patentability issued by KIPO, ch. 1, § 4.1.4 (Jan. 2010). But the same guidance recognizes that a claim involving mathematical or data processing may be statutory if it uses technical devices or technical methods to produce useful, concrete, repeatable technical results. Id.

Korean practice is therefore not hostile to software inventions as such. KIPO’s guidance states that a computer program itself is not a statutory invention, but a data-processing unit operating in association with a computer program, its operating method, and a computer-readable medium carrying the program may be statutory when data processing by the program is specifically executed using hardware. Requirements for Patentability issued by KIPO, ch. 1, § 4.1.8 (Jan. 2010). KIPO’s AI Examination Guide applies the same concept to AI inventions: if information processing by software is concretely realized using hardware resources, the information-processing device, working method, computer-readable medium, and program stored in a medium may constitute statutory inventions. KIPO, Examination Guide in the Artificial Intelligence Field, pt. 1, § 3.1.3 (2023).

This point should not be overstated. Korean eligibility does not always require that the claimed computer, processor, or server itself be specially improved. In everyday examination, eligibility may often be satisfied where the claim concretely recites software-hardware cooperation—an information-processing apparatus, working method, or computer-readable medium/program stored in a medium that performs specific information processing—even when the underlying hardware is conventional. The stricter point in the Supreme Court cases is that the claim, considered as a whole, must not be merely artificial decision-making, human mental activity, economic logic, or information presentation dressed in hardware vocabulary.

III. Korean Case Law on Software, Business Methods, and Mental Activity

The most important Korean software/business-method authority for present purposes is the Supreme Court’s 2007Hu265 decision. In that case, the Court stated that a business-method invention implemented using information technology must involve information processing by software concretely realized using hardware. Daebeobwon [Supreme Court], 2007Hu265, Dec. 24, 2008 (S. Kor.). The Court also emphasized whole-claim analysis: even if part of a claim uses natural law, the claim is not a statutory invention if, considered as a whole, it does not use natural law. Id.

On the facts, the application involved an item server that received consulting requests, analyzed purchase information, and generated lifestyle-planning or consulting results. The Court affirmed rejection because the claim did not disclose how software-based information processing was concretely realized using hardware, and the core of the claim amounted to using a database as a tool to produce artificial decisions or consulting results. Daebeobwon [Supreme Court], 2007Hu265, Dec. 24, 2008 (S. Kor.). For Korean practitioners, 2007Hu265 is a warning that a server, database, or generic computing environment is not automatically enough when the claim’s substance remains a consulting method, human judgment, or artificial decision.

The Court reached a similar negative result in the waste-management/barcode case. There, the claimed invention involved distributing barcode stickers and calendars, affixing barcodes to garbage bags, collecting and sorting garbage, and issuing corrective orders when improperly sorted garbage was detected. Daebeobwon [Supreme Court], 2001Hu3149, May 16, 2003 (S. Kor.). Although the claim involved barcode stickers, garbage bags, and computer barcode reading, the Court held that the steps used those tools merely as instruments for human mental activity, legal/administrative arrangements, and artificial decisions. Id.

The Patent Court’s English translation of the underlying waste-management decision is especially useful for comparative purposes. It explains that a claim using artificial decisions, mathematical formulas, or human mental activities does not fall within the Korean concept of an invention, and that the claim must be assessed as a whole. Teukheobeobwon [Patent Court], 2000Heo5438, Sept. 21, 2001 (S. Kor.). The court found the barcode and computer-reading features insufficient because the hardware and software were used merely as tools and no specific means of using a software-hardware combination was claimed. Id.

By contrast, the Supreme Court has recognized software-related subject matter where software controlled hardware and produced technical operation. In the numerical-control input-format case, the Court held that a software-related invention was not merely human mental activity where the claimed control-input format drove software that controlled a numerical controller, caused machine identification, control, and operation, and produced physical transformation outside the hardware. Daebeobwon [Supreme Court], 97Hu2507, Nov. 30, 2001 (S. Kor.). This is the positive Korean case that most naturally supports U.S.-style arguments about machine control and technological improvement.

The Supreme Court’s authentication/key-exchange decision is also important for modern software and cybersecurity claims. The claimed invention related to mutual public-key authentication and key exchange without a third trusted authority. Daebeobwon [Supreme Court], 2009Hu436, Dec. 23, 2010 (S. Kor.). The Court rejected the argument that the claim failed the statutory-invention requirement merely because it included providing a one-time password through an authenticated P2P channel. The Court construed the authenticated channel as an already-authenticated channel and held that, viewed as a whole, the claim was not purely human mental activity. Id.

Together, these cases define the Korean side of the comparison. A claim directed to human judgment, business administration, legal arrangements, or generic information handling may fail. A claim directed to software concretely cooperating with hardware to control machinery, operate a technical system, or carry out a concrete authentication process may satisfy the Korean threshold. But the threshold is still a Korean threshold; it should not be confused with the U.S. abstract-idea inquiry.

IV. The U.S. Baseline: Statutory Categories, Judicial Exceptions, and Prosecution Guidance

U.S. § 101 begins with statutory categories: process, machine, manufacture, and composition of matter. 35 U.S.C. § 101. For Korean applicants, the first trap is assuming that a claim is safe because it is drafted as a method, system, apparatus, or non-transitory computer-readable medium. Under U.S. law, claim format is only the beginning. A claim may fall within a statutory category and still be patent-ineligible if, under the judicial exceptions to § 101, it is directed to an abstract idea, law of nature, or natural phenomenon. Alice Corp. Pty. Ltd. v. CLS Bank Int’l, 573 U.S. 208, 216–18 (2014); Mayo Collaborative Servs. v. Prometheus Labs., Inc., 566 U.S. 66, 70–73 (2012); AI Visualize, Inc. v. Nuance Commc’ns, Inc., 97 F.4th 1371, 1377–78 (Fed. Cir. 2024).

The Supreme Court’s Alice/Mayo framework asks, first, whether the claims are directed to a patent-ineligible concept and, second, if so, whether the claim elements, individually or as an ordered combination, transform the nature of the claim into a patent-eligible application. Alice, 573 U.S. at 217–18. In the USPTO, that framework is implemented through examination guidance, including Step 2A Prong One, Step 2A Prong Two, and Step 2B. MPEP § 2106. The familiar USPTO abstract-idea groupings—mathematical concepts, certain methods of organizing human activity, and mental processes—are therefore indispensable for prosecution. MPEP § 2106.04(a)(2).

The distinction between judicial law and examination guidance should be explicit. For prosecution, the USPTO’s Step 2A Prong Two framework is the operative vocabulary: when a claim recites a judicial exception, examiners ask whether additional claim elements integrate the exception into a practical application so that the claim is not “directed to” the exception. MPEP §§ 2106.04(d), 2106.05(a); see also 2024 Guidance Update on Patent Subject Matter Eligibility, Including on Artificial Intelligence, 89 Fed. Reg. 58,128, 58,133–34 (July 17, 2024). For litigation, however, the binding law remains Alice/Mayo as interpreted by the Federal Circuit. USPTO guidance and examples are useful prosecution tools, but they are not binding on courts. See In re Rudy, 956 F.3d 1379, 1382–83 (Fed. Cir. 2020); Cleveland Clinic Found. v. True Health Diagnostics LLC, 760 F. App’x 1013, 1020 (Fed. Cir. 2019).

The drafting strategy should therefore satisfy both audiences. It should support a USPTO “practical application” argument during prosecution, while also making the claimed technological improvement visible in the claim language itself for later Federal Circuit-style analysis. This dual audience is especially important for Korean applicants because U.S. claims may be examined under MPEP vocabulary but later tested in litigation under Alice/Mayo and Federal Circuit precedent.

Modern Federal Circuit cases emphasize that the § 101 inquiry is driven by the claimed advance, not by how the applicant characterizes the invention in general terms. The court looks to the claim language, read in light of the specification, to identify what the claim is “directed to.” But the specification cannot rescue a claim by supplying unclaimed technological details. AI Visualize, 97 F.4th at 1378–81; GoTV Streaming, LLC v. Netflix, Inc., 166 F.4th 1053, 1061–66 (Fed. Cir. 2026); Trs. of Columbia Univ. in the City of N.Y. v. Gen Digital Inc., No. 2024-1243, slip op. at 8–14 (Fed. Cir. Mar. 11, 2026). Thus, for U.S. purposes, it is not enough for a Korean-origin specification to describe an implementation that sounds technical if the independent claims themselves merely recite functional results.

Alice remains the central U.S. case for computer-implemented business methods. The claims involved intermediated settlement, and the Supreme Court held that they were directed to the abstract idea of intermediated settlement; merely requiring generic computer implementation did not transform the abstract idea into patent-eligible subject matter. Alice, 573 U.S. at 221–26. The Court stressed that the claims did not improve the functioning of the computer itself or effect an improvement in another technology or technical field. Id. at 225. Bilski is similarly important for business and financial inventions: the Supreme Court held that claims directed to hedging risk in energy markets were directed to an abstract idea. Bilski v. Kappos, 561 U.S. 593, 611–12 (2010).

Recent Federal Circuit cases confirm that this problem remains acute for fintech, gaming, mobile commerce, and regulated transaction systems. In Beteiro, the Federal Circuit held that claims directed to facilitating remote gambling transactions were abstract where they recited detecting information, generating and transmitting a notification, receiving a bet request with location information, determining whether the bet was allowed, and processing the bet. Beteiro, LLC v. DraftKings Inc., 104 F.4th 1350, 1355–59 (Fed. Cir. 2024). The court treated geolocation and regulatory-compliance aspects as legal or business problems rather than technological improvements, and it emphasized that result-focused functional language using generic computer and network technology is generally insufficient. Id. at 1357–59. For Korean fintech applicants, the lesson is direct: a claim framed as payment approval, loan approval, transaction authorization, risk scoring, betting authorization, or regulatory compliance will be vulnerable unless the claim recites a specific technical mechanism rather than merely a computerized commercial decision.

For data-analysis claims, Electric Power Group remains a key Federal Circuit decision. The court held that claims focused on collecting information, analyzing it, and displaying results were directed to an abstract idea. Elec. Power Grp., LLC v. Alstom S.A., 830 F.3d 1350, 1353–56 (Fed. Cir. 2016). Newer Federal Circuit cases have reinforced the same principle. In AI Visualize, the court held that claims involving retrieval, manipulation, and display of medical-scan data were abstract where they recited the result of generating and displaying virtual views without claiming how the system technically achieved the alleged improvement. AI Visualize, 97 F.4th at 1378–82. In GoTV, the court similarly explained that ordinary computer functions—receiving inputs, storing and retrieving information, processing information, displaying information, and transmitting information—do not supply eligibility unless the claim recites a concrete improvement in how those functions are carried out. GoTV, 166 F.4th at 1062–66. These cases are particularly relevant to Korean AI, monitoring, recommendation, image-processing, and analytics applications because many such claims can be reduced by a U.S. examiner or court to “receive data, analyze data, and output a result.”

The Federal Circuit has now made the same point expressly in the machine-learning context. In Recentive Analytics, the court addressed, as a question of first impression, whether claims that merely apply established machine-learning techniques to a new data environment are patent eligible. Recentive Analytics, Inc. v. Fox Corp., 134 F.4th 1205, 1211–15 (Fed. Cir. 2025). The court held that they are not. Id. The claims involved using machine learning to generate event schedules and network maps, but the patents did not claim an improvement to machine learning itself or any specific technological improvement in the underlying computing systems. Id. For Korean AI applicants, Recentive is important because it rejects a common assumption: novelty in the training data, business field, or application environment does not by itself make an AI claim patent-eligible in the United States. A U.S. AI claim should instead identify a concrete technical improvement, such as an improved model architecture, training process, feature transformation, data structure, hardware interaction, network-security action, signal-processing pipeline, or device-control operation.

Cybersecurity and malware-detection claims also require careful treatment. Although computer-security inventions can be eligible, the Federal Circuit has cautioned that merely invoking a security field does not automatically solve § 101. In Columbia University v. Gen Digital, the court held that claims directed to comparing data against a model to detect anomalies were abstract where the claims did not recite the specific technological steps necessary to achieve the alleged security improvement. Trs. of Columbia Univ. in the City of N.Y. v. Gen Digital Inc., No. 2024-1243, slip op. at 12–18 (Fed. Cir. Mar. 11, 2026). The decision is useful for Korean applicants because it shows that even in a technical field such as malware detection, the U.S. inquiry still asks whether the claim recites the how of the technological improvement, not merely the desired security result.

At the same time, U.S. law is not anti-software. The positive line of Federal Circuit authority remains important. In Enfish, the court held that claims directed to a self-referential table for a computer database were not abstract because they were directed to a specific improvement in computer functionality. Enfish, LLC v. Microsoft Corp., 822 F.3d 1327, 1335–39 (Fed. Cir. 2016). Other Federal Circuit decisions have likewise found eligibility, or at least plausibly eligible claims, where the claim recited a specific improvement in computer security, network operation, data structures, packet processing, or device functionality. See, e.g., Finjan, Inc. v. Blue Coat Sys., Inc., 879 F.3d 1299, 1303–05 (Fed. Cir. 2018); Uniloc USA, Inc. v. LG Elecs. USA, Inc., 957 F.3d 1303, 1307–09 (Fed. Cir. 2020); Packet Intelligence LLC v. NetScout Sys., Inc., 965 F.3d 1299, 1309–14 (Fed. Cir. 2020); CardioNet, LLC v. InfoBionic, Inc., 955 F.3d 1358, 1368–72 (Fed. Cir. 2020); CosmoKey Sols. GmbH & Co. KG v. Duo Sec. LLC, 15 F.4th 1091, 1097–99 (Fed. Cir. 2021). The common thread is that the claims did not merely use a computer as a tool. They recited a specific technological mechanism or improvement.

The practical lesson for Korean applicants is therefore sharper than the older Alice/Bilski formulation alone might suggest. A Korean-origin application should not assume that U.S. eligibility follows from claiming a server, processor, memory, terminal, AI model, database, or computer-readable medium. Under current Federal Circuit authority, the claims should identify the technological improvement and recite the specific mechanism that produces it. Otherwise, the U.S. claim may be characterized as a mathematical concept, mental process, business arrangement, regulatory decision, or generic data-processing result, even if the Korean counterpart appears to satisfy the Korean requirement of a technical idea implemented through hardware.

V. The Core Difference: Korean “Hardware Realization” Is Not the Same as U.S. “Practical Application”

The essential comparative point is this: Korea and the United States may identify similar danger signs, but they ask different legal questions.

In Korea, the question is whether the claim, considered as a whole, is a technical idea utilizing laws of nature. In software and business-method cases, that often becomes the practical question whether software information processing is concretely realized using hardware. Daebeobwon [Supreme Court], 2007Hu265, Dec. 24, 2008 (S. Kor.); KIPO, Examination Guide in the Artificial Intelligence Field, pt. 1, § 3.1.3 (2023). Importantly, Korean eligibility may often be satisfied by concrete software-hardware cooperation, even where the hardware is not specially improved. This is one reason Korean applicants may reasonably expect that a computer-implemented software or AI invention is eligible once the claim is drafted as an apparatus, method, or medium with specific software-hardware interaction.

In the United States, the question is different. The claim may recite hardware and still fail if the claimed advance is characterized as a business rule, mental process, mathematical calculation, information-processing result, or generic computer implementation. For prosecution, the applicant will usually argue that any recited judicial exception is integrated into a practical application under MPEP § 2106.04(d). For litigation, the same drafting must support an Alice/Mayo argument that the claims are directed to a technological improvement or contain an inventive concept. Alice, 573 U.S. at 217–18; In re Rudy, 956 F.3d at 1382–83.

This is why Korean eligibility cannot be used as a proxy for U.S. eligibility. A Korean claim reciting “a server comprising a processor configured to receive user information, analyze the user information using an AI model, determine a recommendation, and transmit the recommendation to a user terminal” may appear to involve software-hardware implementation. But in the United States, an examiner or court may characterize the claim as receiving information, performing mathematical or mental analysis, and displaying a result. That characterization places the claim close to Electric Power Group and the MPEP’s abstract-idea groupings. Elec. Power Grp., 830 F.3d at 1353–56; MPEP § 2106.04(a)(2).

The same point applies to computer-readable media. Korean guidance recognizes a computer-readable medium carrying a program as statutory where the data-processing process by the program is specifically executed using hardware. Requirements for Patentability issued by KIPO, ch. 1, § 4.1.8 (Jan. 2010). In the United States, however, a non-transitory computer-readable medium may fall within a statutory category but still fail Alice if the underlying method is an abstract idea implemented on generic computer components. Alice, 573 U.S. at 226–27.

The strongest U.S. approach is therefore not simply to add hardware nouns. The stronger approach is to identify the specific technological problem, the specific technological mechanism, and the specific technological improvement. That is the U.S. bridge from Korean “technical idea” to U.S. “practical application.”

VI. Reframing Korean-Origin Applications for U.S. § 101

For U.S. filing, Korean counsel should begin by identifying the Korean technical idea, but should not stop there. The next step is to ask how a U.S. examiner could characterize the claim at Alice step one or under the USPTO’s Step 2A Prong One. Could the claim be described as a mathematical calculation, a mental judgment, a business interaction, a legal or financial arrangement, a recommendation, a ranking, a prediction, or data collection and display? If so, the application needs a U.S.-specific eligibility strategy.

Because this article is practitioner-oriented, it uses USPTO terms such as Step 2A Prong One, Step 2A Prong Two, and practical application. Those terms are crucial for responding to Office actions. They should not, however, be confused with binding litigation terminology. The same application should also be drafted so that, if the patent is later challenged in court, the claim language itself shows a concrete technological improvement under Alice/Mayo and Federal Circuit law. See In re Rudy, 956 F.3d at 1382–83; Columbia, No. 2024-1243, slip op. at 8–10.

The specification should be drafted to answer a different question from the one that may be sufficient in Korea. It should not merely state that the invention improves convenience, accuracy, efficiency, or user satisfaction. Those are often business or result-oriented benefits. The specification should explain what specific computer, network, database, AI model, sensor, signal-processing system, cybersecurity system, or device operation is improved. The MPEP states that one way to integrate a judicial exception into a practical application is to improve the functioning of a computer or another technology or technical field, and that the specification should provide sufficient technical detail for a person of ordinary skill to recognize the improvement. MPEP §§ 2106.04(d)(1), 2106.05(a).

The claims must also recite the technical mechanism. It is usually insufficient for the specification to describe a technical improvement if the independent claim recites only the result. The USPTO’s 2024 AI eligibility examples illustrate this point for prosecution, while AI Visualize, GoTV, Recentive, and Columbia show the same concern in judicial form. In Example 47, the USPTO treats a broad AI anomaly-detection claim as ineligible where it recites an abstract idea and does not integrate the exception into a practical application, but treats a network-security claim as eligible where the claim applies the AI output to real-time malicious-packet remediation, including dropping malicious packets and blocking future traffic from a source address. 2024 Guidance Update, 89 Fed. Reg. at 58,133–34; USPTO, July 2024 Subject Matter Eligibility Examples 47–49, Example 47.

This example maps well onto Korean practice, but the contrast should be kept sharp. A Korean attorney may be inclined to show that the AI process is concretely implemented by a computer. That may often be enough for Korean eligibility. The U.S. attorney must go further and show that the claim is not merely using AI to classify data. It is using a particular AI process in a technical environment to improve network security through concrete remediation. That is the U.S. eligibility story.

The same conversion should be made for fintech and platform inventions. A claim to receiving loan information, calculating a risk score, determining approval, and displaying a recommendation is vulnerable in the United States because it can be characterized as financial decision-making or a mental process implemented on a generic computer. Alice, 573 U.S. at 219–26; Bilski, 561 U.S. at 611–12; MPEP § 2106.04(a)(2). A stronger U.S. claim would identify a concrete technical mechanism: cryptographic authentication, tokenization, replay prevention, secure API routing, database sharding, reduced authentication latency, intrusion detection, or another computer/network improvement.

The Korean 2009Hu436 authentication case is a useful model. The claim survived the Korean eligibility challenge because, viewed as a whole, it was not merely human mental activity; it involved a concrete authentication/key-exchange process. Daebeobwon [Supreme Court], 2009Hu436, Dec. 23, 2010 (S. Kor.). In the United States, that same type of technology should be framed not as “verifying a user” or “authorizing a transaction,” but as a specific improvement in secure communication, public-key authentication, one-time password delivery, replay resistance, or trusted-channel formation.

VII. Drafting Examples

Consider an AI anomaly-detection invention.

A weak U.S. claim might recite:

A computer-implemented method comprising receiving data, applying an artificial intelligence model to the data, determining whether an anomaly exists, and outputting anomaly information.

This claim may be acceptable in some Korean contexts if the software-hardware implementation is concretely described. But in U.S. practice, it is vulnerable because the claim can be characterized as collecting information, analyzing it using mathematics, and outputting a result. Elec. Power Grp., 830 F.3d at 1353–56; MPEP § 2106.04(a)(2).

A stronger U.S. version would recite:

A computer-implemented method comprising receiving packet-flow data from network sensors, generating a time-correlated feature vector representing packet behavior, applying the feature vector to a trained neural network configured to detect intrusion-pattern anomalies, identifying a malicious network session, detecting a source address associated with the malicious network session in real time, dropping packets associated with the malicious network session, and blocking future traffic from the source address.

The point is not that this exact language is always eligible. The point is that the claim now has a U.S. eligibility story. It identifies a technical environment, a specific data source, a technical analysis, and a technical remediation step. As prosecution guidance, it is closer to the USPTO’s eligible AI network-security example than to a generic data-analysis claim. USPTO, July 2024 Subject Matter Eligibility Examples 47–49, Example 47. As litigation drafting, the same language is also better positioned because it recites concrete action rather than merely claiming the result of “detecting an anomaly.”

Consider next a fintech authentication invention.

A weak U.S. claim might recite:

A method comprising receiving transaction information, determining whether the transaction is authorized, and transmitting an authorization result.

This claim invites an Alice rejection because it may be characterized as a commercial authorization decision implemented on a generic computer. Alice, 573 U.S. at 219–26.

A stronger U.S. claim would recite:

A method comprising generating a transaction-specific limited-use token, embedding the token in a machine-readable authentication link, receiving a response message containing token-linked transaction data, verifying the response message using domain-authentication data, determining whether the token has expired or has been replayed, updating a replay-prevention data structure, and granting network access to a transaction resource only when the token verification and replay-prevention checks are satisfied.

That claim may still face § 101 scrutiny, but the U.S. framing is stronger. The claimed advance is no longer merely “authorizing payment.” It is a particular computer-network authentication mechanism. The claim is also better aligned with the claim/specification warning stated at the outset: the technical mechanism appears in the claim, not only in the specification.

VIII. Conclusion

Korean and U.S. eligibility law overlap, but they are not interchangeable. Korean law asks whether the claimed subject matter is a technical idea utilizing laws of nature and, in software-related cases, whether software information processing is concretely realized using hardware. In practical KIPO examination, concrete software-hardware cooperation may often be enough, even where the hardware itself is conventional. U.S. law asks a different question: whether the claim falls within a statutory category and avoids, or sufficiently applies, judicial exceptions such as abstract ideas, laws of nature, and natural phenomena.

The Korean cases provide useful guidance for identifying danger zones. Claims based on human judgment, artificial decisions, business rules, legal arrangements, or generic use of computers are vulnerable in Korea and in the United States, though for different doctrinal reasons. Claims involving concrete machine control, authentication architecture, network security, database improvement, signal processing, or physical-device operation are more likely to support both Korean eligibility and U.S. § 101 eligibility.

For Korean applicants, the practical lesson is straightforward: do not treat U.S. filing as an English translation exercise. A U.S.-ready application should convert the Korean technical idea into a U.S. § 101 narrative. The application should explain the technical problem, disclose the technical mechanism, identify the technological improvement, and claim the limitations that produce that improvement. For prosecution, that supports a USPTO “practical application” argument. For litigation, it helps show under Alice/Mayo and Federal Circuit law that the claim itself is directed to a concrete technological improvement rather than an abstract idea implemented on a generic computer.