Home > AI Litigation and Licensing Business people chatting at corporate event, meeting in corporate bar by large window, young man using laptop AI PATENT INTELLIGENCE The new shape of AI patent risk Artificial intelligence patenting has reached meaningful scale, but the larger signal is structural. Ownership is fragmented, litigation is becoming more active in mature domains, and foundational AI is now important enough to inspire shared licensing models such as SAIL. Explore the analysis Discuss your portfolio position Risk Landscape Fragmentation Litigation PAEs and R&D Player SAIL Strategic Implementation AT A GLANCE The AI landscape is large, distributed, and moving into a more contested phase. 790K Patent families across four core AI domains Machine learning, natural language processing, computer vision, and generative AI. 2.5xMachine learning litigation growth Litigation events rose more than 2.5 times from 2024 to 2025. 0.05%Share held by PAEs Small by count, but concentrated enough to create targeted enforcement risk. ~2,800Patent families aligned with SAIL scope A focused subset of foundational AI technologies rather than the whole THE AI PATENT BASELINE Why the risk picture is changing now AI is no longer a single monolithic category. It is a broad technology stack spanning machine learning, natural language processing, computer vision, and generative AI, each with different levels of maturity, ownership patterns, and enforcement signals. Together, these four domains already represent about 790,000 patent families, which means exposure can expand quickly as organizations deploy AI across products, processes, and infrastructure. The analysis uses AI-specific technology classifiers built with LexisNexis® Classification to separate machine learning, natural language processing, computer vision, and generative AI into comparable domains. 393KMachine learning The largest domain in the analysis, with the clearest recent escalation in litigation activity. 219KComputer vision A mature domain with an earlier litigation wave that has settled into steadier activity. 153KNatural language processing A substantial domain with sustained litigation pressure and notable PAE exposure. 23KGenerative AI Still the smallest by volume, but increasingly important because foundational technologies scale across many use cases. Machine learning is the widest and most active field Data-driven models that improve through training underpin decision support, optimization, and prediction across industries. Largest domain and strongest recent litigation acceleration. Natural language processing remains strategically exposed Technologies for understanding and generating language already support large product surfaces and service workflows. Consistently elevated litigation and the highest PAE concentration. Computer vision is mature enough to show historical cycles Image and video interpretation has already experienced earlier litigation waves, giving innovators a view of how disputes can evolve over time. Earlier surge, then steadier post-peak activity. Generative AI is still early, but not peripheral Foundational models, training methods, evaluation layers, and integration tooling can influence an entire ecosystem, not just one application. Smallest domain today, but central to future platform risk. OWNERSHIP STRUCTURE A fragmented market can still produce concentrated risk Even when top owners hold a relatively modest share of patents, enforcement risk can still come from many directions because the remaining portfolios are spread across a long tail of owners. Top 10 ownership share by domain Global figures and views excluding China tell different stories. Removing China often raises concentration in machine learning, natural language processing, and computer vision, while generative AI becomes more distributed. Machine learningGlobal top 10 share: 12.2% | Excluding China: 16.7% Natural language processingGlobal top 10 share: 12.0% | Excluding China: 26.1% Computer visionGlobal top 10 share: 13.1% | Excluding China: 22.5% Generative AIGlobal top 10 share: 25.5% | Excluding China: 9.7% What the comparison suggests China materially shapes the apparent fragmentation of machine learning, natural language processing, and computer vision when global patenting is viewed as a whole. Generative AI behaves differently. The higher global top 10 share suggests concentrated early leadership, while the lower share excluding China points to a broader set of non-China players. For innovators, fragmentation does not reduce risk. It often makes exposure harder to map because the relevant patents can sit outside the best-known portfolios. GLOBAL MACHINE LEARNING Broad base, rising pressure Machine learning remains deeply fragmented globally, but concentration increases once China is removed. That mix of scale and dispersed ownership helps explain why litigation activity can rise quickly. NATURAL LANGUAGE PROCESSING Fragmentation hides important control points The global top 10 share is modest, yet the ex-China view rises sharply. That suggests a field where a relatively small number of non-China portfolios may carry disproportionate strategic weight. COMPUTER VISION Maturity and breadth coexist Computer vision combines a large installed base of patents with a long tail of owners. The result is a domain where historical precedent and ongoing commercial relevance overlap. GENERATIVE AI Early concentration sits at the core layer Generative AI is more concentrated globally than the other domains, which reinforces why foundational technologies, model tooling, and shared licensing have become focal points so early. Ownership distribution This comparison shows how concentration changes once the largest national filing base is removed and why the excluding-China view sharpens strategic control points. Machine learning Natural Language processing Computer Vison Generative AI ML: Top Owners & Portfolio Share (Excl.China) NLP: Top Owners & Portfolio Share (Excl.China) CV: Top Owners & Portfolio Share (Excl.China) Gen AI: Top Owners & Portfolio Share (Excl.China) DISPUTE ACTIVITY Litigation is not evenly distributed across AI Different AI domains are showing different maturity patterns. That makes it more useful to read litigation as a set of domain-specific signals rather than one aggregated trend line. MACHINE LEARNING Recent acceleration stands out Machine learning has the largest total landscape and 158 litigation events in the analyzed set. Events rose more than 2.5 times from 2024 to 2025, which points to an enforcement environment that may stay elevated as adoption spreads further into enterprise systems and infrastructure. NATURAL LANGUAGE PROCESSING Litigation pressure is broad and persistent Natural language processing recorded 880 litigation events, the highest figure in the analysis. After rising again around 2020, activity has stayed elevated, suggesting a domain where foundational claims and downstream product use can continue to intersect. COMPUTER VISION An earlier wave offers a cautionary pattern Computer vision logged 593 litigation events. The pattern shows a sharp surge around 2012 and 2013, followed by a decline and then a steadier level from 2019 onward. That cycle matters because it shows how quickly a growing technical field can become litigated once commercial applications broaden. GENERATIVE AI Low event counts should not be mistaken for low future risk Generative AI shows only 14 litigation events in the current data set. That reflects a smaller and newer landscape, not immunity. If foundational patents keep accumulating around models, training, evaluation, and integration, the domain may follow a path already seen in older AI fields. The practical read The litigation story is not that every AI domain is equally dangerous today. It is that maturity patterns matter. Machine learning and natural language processing already show active pressure. Computer vision offers a historical precedent. Generative AI is still early, but it sits close to the foundational layer where future disputes can have wide impact. NAMED PARTIES The plaintiff and defendant record is easiest to understand in context Each litigation trend becomes more concrete when viewed alongside the named plaintiffs and defendants. The tables below show how the mix of parties changes by domain, and how quickly foundational claims travel into large operating companies. Machine learning Natural language processing Computer vision Gererative AI Machine learning Recent acceleration is paired with a party list that mixes assertion-oriented entities, chip-linked transfers, and major platform defendants. Machine learning litigationThis view combines the litigation trend with named plaintiffs and defendants for a single-page read of pace, scale, and exposure. The combination reinforces how quickly machine learning claims can move from portfolio ownership into broad commercial exposure. Plaintiffs: Rondevoo Technologies, Fractal Networks, Tesla Defendants: Google, Microsoft, Intel Count of litigation events in ML Top Plaintiffs Rondevoo Technologies LLC Modern Telecom Systems LLC Mobile Health Innovative Solutions LLC Bush Seismic Technologies LLC Patent Armory Inc. Fractal Networks LLC Dental Monitoring SAS Webroot Inc. Fred Bassali Bally Technologies Inc. Tesla Inc. Tempus AI Inc. Metarail Inc. Health Discovery Corporation All Terminal Services LLC Top Defendants Google LLC Tesla Inc. Align Technology Inc. Johnson Controls Inc. Microsoft Corporation Business Intelligence Systems Solutions Inc. Guardant Health Inc. Intel Corporation Roboflow Inc. Asustek Computer Inc. Autonomous Devices Inc. H2O.AI Inc. Katherine K. Vidal MicroStrategy Inc. Mindshare Medical Inc. Natural language processing Natural language processing remains the most established litigation field in the study, with a deep record of repeat defendants and high-value enforcement targets. The table makes clear that foundational language claims already reach widely distributed product and cloud ecosystems. Plaintiffs: LINFO IP, Cedar Lane Technologies Defendants: Google, Apple, Amazon, Microsoft Count of litigation events in NLP Top Plaintiffs LINFO IP, LLC Cedar Lane Technologies, Inc. IPA Technologies, Inc. Hitel Technologies, LLC Spider Search Analytics, LLC R2 Solutions, LLC Ultratec, Inc. Semantic Search Technologies, LLC Verna IP Holdings, LLC Google LLC DataCloud Technologies, LLC Selene Communication Technologies, LLC UnoWeb Virtual, LLC Milestone IP, LLC Apple, Inc. Top Defendants Google LLC Apple, Inc. Amazon.com, Inc. Microsoft Corporation CaptionCall, LLC Samsung Electronics Co., Ltd. [unknown] Sonos, Inc. IPA Technologies Inc. Sorenson Communications, Inc. MotionPoint Corporation Facebook, Inc. International Trade Commission LG Electronics, Inc. Taasera Licensing LLC Natural language processing litigationThis view shows the highest litigation event count in the analysis alongside a defendant set dominated by the largest software and platform players. Computer vision Computer vision offers the clearest historical precedent in the analysis, with a matured litigation pattern and a defendant base spanning several industries. The table format shows how computer vision disputes traveled well beyond a single product category. Plaintiffs: Unified Messaging Solutions, Intellectual Ventures Defendants: BMW, Samsung, Apple Count of litigation events in Computer vision Top Plaintiffs Unified Messaging Solutions, LLC Scanning Technologies Innovations, LLC Canatelo, LLC Pebble Tide LLC Express Card Systems, LLC Stragent LLC West View Research, LLC Andra Group, LP United Services Automobile Association Eagle View Technologies, Inc. NorthStar Systems LLC Content Aware, LLC Flexiworld Technologies, Inc. Intellectual Ventures I LLC Wapp Tech Limited Partnership Top Defendants Smart Mobile Technologies LLC PNC Bank National Association Camtek Ltd. Google LLC Facebook, Inc. United Services Automobile Association BMW of North America, LLC Bayerische Motoren Werke AG International Trade Commission Xactware Solutions, Inc. Apple, Inc. MapleBear Inc. Motive Technologies Inc. Roku, Inc. Samsung Electronics Co., Ltd. Computer vision litigationThe earlier dispute wave is visible in the chart, while the party lists show how claims moved into automotive, electronics, and consumer-device contexts. Generative AI The layout makes the early but distinct risk pattern visible without overstating its present scale. Generative AI is the newest litigation field in the analysis, but the party list already shows how quickly model-linked claims can attach to emerging application providers. Defendants: Canva, Eleven Labs, Synthesia Plaintiffs: Cedar Lane, Sanas.AI Count of litigation events in GenAI Top Plaintiffs Cedar Lane Technologies, Inc. Balor Audio, LLC Sanas.AI Inc. Infinity Cube Ltd. Top Defendants Ableton, Inc. Avid Technology, Inc. Bandlab U.S.A., Inc. Canva US, Inc. Descript, Inc. Eleven Labs Inc. Krisp Technologies, Inc. Mangolytics, Inc. Mark of the Unicorn, Inc. Neosapience, Inc. Resemble AI Inc. Speechify, Inc. Synthesia Ltd Tracktion Software Corp. Generative AI litigationThe current event count is still small, but the landscape already shows an identifiable set of named parties around synthetic media and model-enabled workflows. Explore litigation activities and top plaintiffs and defendants in your technology fields Talk to our experts PAE DYNAMICS A small PAE footprint can still carry strategic weight Patent assertion entities hold a tiny share of total AI patent families, but their portfolios are concentrated enough to support targeted litigation in strategically chosen domains. SCALE PAEs hold a small slice of the whole 393 Active patent families are identified as PAE-held across the four AI domains, out of a total landscape of about 790,000 families. CONCENTRATION Ownership is tightly concentrated 80% The top 15 PAEs hold about 80 percent of PAE-owned AI patents, and the top five control roughly 68 percent. EXPOSURE Natural language processing stands out 239 Natural language processing accounts for the largest share of PAE-owned families in the analysis and shows active disputes involving major technology companies. Natural language processing is the sharpest PAE signal Dialect LLC holds a small but high-impact portfolio and has pursued cases against Amazon, Microsoft, Salesforce, and Meta. VB Assets also traces back to Voicebox-originated patents and has been involved in disputes with Amazon, Apple, and SoundHound AI. The lesson is not just volume. Small portfolios with focused claims can still be influential. Machine learning and computer vision show targeted acquisitions Lodestar Licensing appears in both machine learning and computer vision through Micron-originated patents with chip-related relevance. Cloud Byte and Fractal Networks illustrate how transferred portfolios can create new assertion paths. Computer vision also includes newer enforcement entities such as Artificial Intelligence Industry Association. Software orientation matters The pattern points to a strong skew toward software-related patents across PAEs. That raises the relevance of claims that travel easily across applications, services, and enterprise workflows. Generative AI exposure is still low by count today, but early-stage concentration can change quickly as portfolios consolidate. Why this matters for operating teams PAE exposure in AI is not yet a scale story. It is a selectivity story. A relatively small number of entities can still create pressure if they control high-impact patents in technologies that sit inside widely adopted software and model workflows. 72 Machine learning PAE-held families in the analysis. 239 Natural language processing PAE-held families, the largest PAE exposure among the four domains. 76 Computer vision PAE-held families, with newer assertion entrants joining the field. 7 Generative AI PAE-held families today, which is low, but consistent with an early-stage field. PAE ownership overview This summary view shows how a small group of patent assertion entities controls most PAE-held AI families despite their modest share of the total landscape. Learn more about the PAE dynamics in your core technology field Reach out to our experts How licensing-focused R&D players shape AI patent dynamics Adeia and InterDigital show how monetization power can sit with portfolio builders rather than end-user product companies. Adeia shows a broad portfolio with a notable concentration in natural language processing, while InterDigital is more machine learning-led. Their presence matters because meaningful AI leverage can come from organizations whose core role is portfolio development and monetization rather than end-user product delivery. SHARED LICENSING SAIL points to a new response at the foundational layer The Shared AI License initiative is not designed to cover all AI. It focuses on the parts of the stack where patent friction could slow an entire ecosystem rather than a single application. SAIL is structured as a defensive patent commons in which members cross-license certain AI patents on a royalty-free basis. The practical signal is important: participants are trying to reduce litigation around the technical building blocks of foundation models, rather than around end-user product features. What falls inside scope Foundation models Training, fine-tuning, and adaptation methods Testing, verification, validation, and monitoring Integration layers, APIs, and programming frameworks Safety mechanisms, controls, and oversight capabilities What falls outside scope End-user applications built on foundation models Domain-specific implementations at the application layer Product features intended primarily for user workflows Hardware infrastructure and physical architectures General downstream use cases in sectors such as design, finance, or life sciences PORTFOLIO ANALYSIS Translating the SAIL agreement into a patent search produced about 2,800 patent families aligned with the initiative’s defined scope, which suggests a focused but strategically important slice of the broader AI market. SAIL technology cluster view This chart maps the initiative’s technology center of gravity around foundational AI capabilities rather than downstream application categories. Portfolio size and impact within scope This comparison shows why portfolio relevance and LexisNexis Patent Asset Index together give a more useful signal than raw patent volume alone. Member structure is deliberately selective Founding and board members include Anthropic, IBM, Meta, Microsoft, and Genentech. Additional members include Block, Figma, Canon, and observers such as eBay and TD Bank Group. Portfolio alignment differs widely by participant Some members contribute a small portion of their AI portfolios to foundational technologies, while others align much more heavily. Anthropic stands out as having an especially high share of patents that fit the SAIL criteria. Portfolio impact was assessed using the LexisNexis Patent Asset Index, combining technology relevance and market coverage. Applied AI is not the same as foundational AI Block, Figma, and Genentech show meaningful AI activity, but the analysis places most of it at the application layer. Their patents largely support financial workflows, design experiences, or biomedical use cases rather than the enabling technologies SAIL is built to cover. What SAIL really signals SAIL suggests that the highest perceived patent risk sits in shared technical infrastructure such as model development, evaluation, integration, and safety. That is where a single patent can affect many companies at once, and where cooperative licensing starts to make strategic sense. Founding and core members These participants sit inside the core licensing network and shape the initiative’s direction as well as its practical defensive value. Anthropic | IBM | Meta | Microsoft | Genentech Additional members Additional members contribute patents and receive licenses, but the analysis indicates their alignment with foundational AI can vary widely by portfolio composition. Block | Figma Observers and potential future participants Observers will monitor how the initiative evolves and whether joining becomes strategically attractive. eBay | TD Bank Group Share of SAIL-aligned patents across portfolios This chart makes the portfolio concentration point visible, especially the much higher alignment shown for Anthropic relative to the rest of the current group. Potential additional U.S. participants The screening table shows how the analysis extends beyond current members and identifies other U.S. companies with meaningful foundational AI relevance. Geographic comparison in foundational AI The country-level comparison places the combined SAIL portfolio alongside large regional clusters across the United States, China, Japan, Korea, and Europe. Applied AI that sits outside the framework This visual shows why several current participants have visible AI portfolios that still fall mostly outside SAIL’s foundational scope. Why some members sit mostly outside scope This chart makes the portfolio concentration point visible, especially the much higher alignment shown for Anthropic relative to the rest of the current group. How the analysis extends beyond current members The screening table shows how the analysis extends beyond current members and identifies other U.S. companies with meaningful foundational AI relevance. Looking to evaluate your portfolio or a competitor’s position? Get in touch with our experts WHAT THIS MEANS FOR INNOVATORS Three strategic implications stand out The most resilient AI IP strategies start with visibility, then layer in forward-looking monitoring and a realistic view of where collaboration can reduce structural risk. IMPLICATION 1 Visibility into foundational technologies matters more than ever If you cannot see which patents cover core AI capabilities, ownership changes, and technology boundaries, it becomes harder to assess exposure before products and partnerships scale. IMPLICATION 2 Litigation risk is likely to widen with adoption Companies do not need to build models in-house to become exposed. As AI becomes embedded across tools, services, and workflows, more organizations may find themselves inside the path of assertion activity. IMPLICATION 3 Collaboration models are becoming part of IP strategy SAIL shows that for foundational AI, portfolio strategy is no longer only about exclusion or bilateral licensing. It can also include shared defensive structures that lower friction across an ecosystem. GET IN TOUCH Gain clarity on how your portfolio is positioned across technology domains LexisNexis Intellectual Property Solutions helps innovators assess portfolio relevance, ownership dynamics, and technology positioning with high-quality patent data, analytics, and domain-specific methodology. Use the form to reach our team about portfolio analysis, litigation monitoring, foundational AI risk, or SAIL-related questions.
