Differential Privacy Accounting Patents: Leaders & Trends 2026
Filing growth compares 2021 (38 records) with 2024 (77) — a three-year span. 2024 is the most recent year we treat as complete: publication lags filing by roughly 18 months, so 2025 onwards are still filling in and any growth rate that ends there would understate the field. Top-5 share is the combined record count of the five largest assignees divided by all 502 records in scope (CR5), not by the ranked leaders only.
What differential privacy accounting patents actually cover
Differential privacy accounting is the discipline of tracking and bounding how much information a system leaks each time it queries a dataset, then enforcing a cumulative privacy budget across repeated queries or training steps. In patent filings this shows up less as abstract cryptographic theory and more as applied machinery bolted onto existing data pipelines: federated learning systems that inject calibrated noise before aggregation, hyperparameter tuners that adjust noise scale against a fixed budget, and monitoring layers that halt a process once a privacy loss threshold is crossed.
The 502 records in scope span 2015 through the middle of 2026, with filings concentrated overwhelmingly in general computing and AI-adjacent classes rather than in dedicated cryptography classifications. That composition signals a field still being built inside machine-learning infrastructure, not as a standalone cryptographic primitive.
Let an AI agent run this analysis on your own technology
Pick a task. Every answer cites the patents behind it.
Filing trend and technology composition
Two views of the same 502-record dataset: how filing volume has moved year over year, and which IPC subclasses carry the claim density.
Filing trend, 2017-2026
Annual filings rose from 14 in 2017 to a peak of 152 in 2025, with 41 recorded so far in 2026. Because publication lags filing by roughly 18 months, the most recent one to two years understate true filing activity. The cleanest read of momentum is the complete 2021-2024 window: filings rose from 38 to 77, a +103% increase.
Technology composition by IPC subclass
G06F (electric digital data processing) appears on 82.5% of the 502 records and G06N (AI-based computing) on 47.6%, confirming that most differential privacy accounting claims are written as extensions to general computing and machine-learning systems rather than as dedicated security primitives. H04L (digital transmission, 22.3%) and G06Q (business/admin processing, 12.7%) trail well behind, and healthcare informatics (G16H, 4.2%) and image recognition (G06V, 2.8%) remain minor. Because records can carry multiple classes, these shares add to more than 100% and should be read against the 502-record total, not against each other.
Shares are the percentage of the 502 records in scope. A patent can carry several IPC classes, so the shares add up to more than 100%.
Go deeper on Privacy-Enhancing Technologies: Differential Privacy Accounting Patent Landscape with Eureka
This page is one run against one query. Ask Eureka your own question about privacy-enhancing technologies: differential privacy accounting patent landscape and every answer comes back with the patent numbers behind it.
Try EurekaA representative recent filing
Dynamic smart contract security and verification system using capsule networks, autoencoders, and generative adversarial networks
The filing combines an autoencoder for preprocessing smart contract code, a capsule network for capturing hierarchical dependencies within that code, and a GAN that generates routing coefficients to sharpen the capsule network's analysis, all deployed on a blockchain platform for continuous monitoring of smart contracts.Assignee and publication details are rendered alongside this entry.
| # | Publication no. | Patent title | Citations |
|---|---|---|---|
| 1 | CN107368752A | 一种基于生成式对抗网络的深度差分隐私保护方法 | 184 |
| 2 | US20230068386A1 | Systems and methods for distributed learning for wireless edge dynamics | 109 |
| 3 | WO2021158313A1 | Systems and methods for distributed learning for wireless edge dynamics | 90 |
| 4 | CN113591145A | 基于差分隐私和量化的联邦学习全局模型训练方法 | 69 |
| 5 | US20210216902A1 | Hyperparameter determination for a differentially private federated learning process | 61 |
| 6 | CN111091199A | 一种基于差分隐私的联邦学习方法、装置及存储介质 | 50 |
| 7 | CN114297722A | 一种基于区块链的隐私保护异步联邦共享方法及系统 | 40 |
| 8 | US20250285467A1 | Deep learning-based facial recognition system with privacy-preserving features | 37 |
| 9 | CN110874488A | 一种基于混合差分隐私的流数据频数统计方法、装置、系统及存储介质 | 35 |
| 10 | CN109784091A | 一种融合差分隐私GAN和PATE模型的表格数据隐私保护方法 | 35 |
Citation counts favour older records simply because they have had longer to accumulate citations inside the searched corpus; treat them as a signal of influence, not of current importance.
Patent titles are shown in the language they were filed in, not translated, so that each record stays verifiable against the original filing — a translated title will not match in Eureka or in any national register. Each row carries its publication number; clicking a row searches Eureka by that number.
Put your own technology through the same analysis
Eureka on the web
When you want the answer in the next five minutes.
The agent works the prompt against patents and technical literature, citing every source.
Run your analysis now →MCP server & REST API
When it has to run inside your own pipeline.
Patent search, landscape analysis and assignee resolution as MCP tools. Drop them into any agent framework, or call REST directly.
Browse MCP servers →What the concentration and growth figures mean for filing strategy
Three figures from this dataset matter more than the raw counts: how spread out ownership is, how fast the field moved before the data went quiet, and where the claim space is already crowded.
No single assignee controls the field
The leading assignee holds 15 records and the fifth-ranked holder just 8; combined, the top 5 account for only 12.0% of all 502 records in scope, and the top 10 for 18.7%. That is a long tail, not a walled garden — most of the ranked leaders hold single-digit counts, which means freedom-to-operate analysis has to look past the household names into a wide field of smaller filers.
Filing activity roughly doubled in three years
Annual filings grew from 38 in 2021 to 77 in 2024, a +103% increase over the last three-year window the data can treat as complete. 2025's count of 152 and 2026's partial 41 suggest the trend kept climbing, but publication lag means those two years will keep filling in and should not yet be read as a peak or a plateau.
Claims sit inside general computing, not dedicated crypto classes
G06F and G06N between them touch the large majority of records (82.5% and 47.6% respectively), meaning most differential privacy accounting claims are drafted as features of computing or AI systems rather than as freestanding cryptographic methods. Filers looking for a cleaner classification lane may find less crowding in G16H (4.2%) or G06V (2.8%).
Filing activity is centred on one receiving office
China accounts for 332 of the 502 records, with India (60) and the United States (58) a distant second and third, and Europe, WIPO and Canada each in single or low double digits. A patent strategy built only around US or EPO filings will miss the majority of documented activity in this field.
Eureka can read the same corpus for gaps instead of for coverage: under-claimed branches adjacent to privacy-enhancing technologies: differential privacy accounting patent landscape, with the prior art for and against each one.
Where to take this analysis next
The dataset points to a field with low ownership concentration and rising filing volume, which changes what due diligence and freedom-to-operate work should focus on next.
Map the long tail of filers
With the top 10 assignees holding under a fifth of all records, a full competitive picture requires screening well beyond the named leaders into the smaller, single- or double-digit filers.
Explore the assignee rankingTrack the 2024-2026 filing window closely
Because publication lag understates the most recent years, revisit filing counts for 2025 and 2026 periodically as more records surface, rather than treating today's snapshot as final.
Monitor filing trends in EurekaCheck classification overlap before drafting
High density in G06F and G06N means new claims drafted purely as generic computing features face crowded prior art; searching adjacent classes with lower density may reveal cleaner claim space.
Run a classification search in EurekaFrequently asked questions
The assignee ranking spans 100 companies drawn from 502 records, with the leader holding 15 records and the tenth-placed holder only 6. No single company dominates: the top 5 combined hold just 12.0% of all records in scope, and the top 10 hold 18.7%. This is a fragmented field with a long tail of filers each holding a handful of records, so competitive monitoring needs to look well past the largest few names.
Using the last three-year window the data can treat as complete, filings grew from 38 in 2021 to 77 in 2024, a +103% increase. Counts continued rising through 2025 (152) and into 2026 (41 so far), but because publication typically lags actual filing by around 18 months, those two most recent years are still filling in and should not be read as the field slowing down. The clean, comparable growth signal is the 2021-2024 figure.
G06F, the electric digital data processing class, appears on 82.5% of the 502 records in scope, followed by G06N (AI-based computing) at 47.6%. This indicates that most patent claims in this space are framed as features added to general computing or machine-learning systems, rather than as standalone cryptographic inventions. Smaller classes such as G16H (healthcare informatics, 4.2%) and G06V (image recognition, 2.8%) show far less claim density and may offer more room to file distinctively.
China's patent office receives by far the most filings in this dataset, 332 of the 502 records, followed by India with 60 and the United States with 58. Europe (EPO), WIPO/PCT filings and Canada each account for smaller counts. A filing or freedom-to-operate strategy focused only on US or European patents would miss the large majority of documented activity in this field.
It is fragmented. Across the 502 records and 100 ranked assignees, the leader holds only 15 records, and combined the top 5 assignees account for just 12.0% of all records in scope. That pattern, a modest leader plus a long tail of small filers, means large incumbents have not locked up the core claim space, leaving room for new entrants but also making comprehensive prior-art clearance more labour-intensive.
Research Privacy-Enhancing Technologies: Differential Privacy Accounting Patent Landscape in depth with Eureka
Go past this page: query the whole privacy-enhancing technologies: differential privacy accounting patent landscape corpus yourself, in your own scope.
Every answer comes back with patent numbers you can open.
Disclaimer. This page is generated from Patsnap Eureka data drawn from a limited snapshot of global patent and scientific-literature records, and is provided for general information and reference only.
Patent data carries inherent limitations: recent filings (typically the most recent 18–24 months) are under-counted due to standard publication lag; counts may be reported at either a patent-family or a patent-record basis and are not always directly comparable; classification, applicant-name, and citation data may contain errors, duplicates, or omissions; and the underlying search query defines and constrains the scope shown. As a result, the analysis may be incomplete or inaccurate and may not reflect the full technology landscape.
Nothing on this page constitutes an exhaustive prior-art, novelty, freedom-to-operate, or validity search, nor does it constitute legal, financial, investment, or professional advice, and it should not be relied upon as such. Any patent, commercial, or strategic decision should be verified independently and reviewed with qualified patent, legal, and domain professionals. Patsnap makes no warranties, express or implied, as to the accuracy, completeness, or fitness for any particular purpose of the information presented.
Machine translation. Assignee and organisation names originally recorded in Chinese, Japanese or Korean have been rendered into English by an AI translation step so that the tables stay readable. These renderings are best-effort and may not match a company’s registered English name; the original name is what the underlying patent record carries, and it is what any Eureka query launched from this page uses.