PERSPECTIVES
-
The problems trademark attorneys are facing are well-documented, from the surge in trademark applications in 2020 and 2021 that resulted in a backlog of 544,000 unexamined classes at the USPTO to trade in fake goods now representing a $509-billion-dollar industry. Online trademark infringement is a growing problem, with the high volume of occurrence being one of the greatest obstacles in pursuing legal action1,2.
There are many intelligent and thoughtful discussions taking place on how to address these challenges, and we’ve observed that many of the written pieces conclude with thoughts about how, in conjunction with other efforts, frontier technologies like AI may be able to assist.
We hope that by offering an overview of the existing and emerging solutions in the trademark law and brand protection ecosystem, we can enable IP professionals to assist their clients to the best of their abilities.
1. “Search & Watch” vs. “Brand Protection”
A Historical Divide
Up through the late 2010s, most of the technology for trademark clearance and enforcement had been segmented into two distinct groups of solutions:
- Search & Watch
- Brand Protection
“Search & Watch” solutions offered trademark clearance search services and the ability to watch or monitor new trademark applications for potential conflicts. These services are incredibly helpful in helping attorneys to secure and maintain the relevant IP rights for their clients, but these solutions don’t assist with protecting trademarks in use. As a result, a new category of “Brand Protection” tools began to emerge to fill this gap.
Table 1. IP technology vendors' founding focus
From the start, most companies focused on only one of these two verticals, although that changed quickly in the late 2010s with a series of rapid acquisitions.
Target Markets
Generally speaking, “Search & Watch” solutions have been marketed to all lawyers and law firms, while Brand Protection tools have historically been offered directly to brand owners or their in-house counsel, not to law firms.
While we can only speculate, we believe that there are two possible reasons for this divide. First, it could be a purely commercial decision. There are more brand owners than there are IP lawyers who work in law firms, which could potentially make targeting this market more lucrative. There is also the possibility that this is a function of cost. If pricing is structured on a per trademark model, the costs would very quickly skyrocket for lawyers who manage trademark portfolios of multiple clients, not just single brands.
What we do know for sure is that many law firms – even sophisticated litigation practices – are allocating human resources to manually searching for possible instances of infringement online and recording the evidence with screenshots and saved links when they would much rather employ some forms of automation.
However, as startups have entered the space and acquisitions have accelerated in recent years, the general trends are starting to change as more and more vendors seek to offer more comprehensive solutions to better help their clients.
2. Big players have big appetites: 28 acquisitions and related activities over 5 years
Between 2017 and June 2022, three of the biggest players in the IP ecosystem – Clarivate, Corsearch, and Questel – were involved in 24 acquisitions, 2 partnerships, 1 collaboration, and 1 investment.
These acquisitions speak to the providers’ ambitions of becoming comprehensive, end-to-end providers in the IP space, with each company having a slightly different take on what that means; Questel’s acquisitions were mostly focused on expanding their patent and portfolio offerings, Clarivate’s acquisitions were mostly focused on the data and analytics space, and Corsearch’s acquisitions were focused on making gains in the online brand and content protection space.
3. Brand protection throughout the entire trademark lifecycle
Corsearch alone acquired 6 companies and partnered with another 2 in less than 3 years, reflecting a broader industry trend of Search & Watch providers starting to enter the Brand Protection space.
Other vendors have also taken steps to offer more comprehensive services. For example, Markify quietly entered the eCommerce infringement detection space following their acquisition by Questel earlier this year; however, their monitoring services are text-only at this time.
Table 2. Search & Watch providers start to enter the Brand Protection space
Then in 2022, a new vendor emerged from the Silicon Valley called Huski. Huski was founded by engineers who are image recognition experts who abandoned working on self-driving cars to join an IP litigator in building a new solution for trademark and brand protection. From the start, Huski offered both Search & Watch and Brand Protection services, as well as native image recognition technology.
4. A continued focus on image recognition technology
Image recognition technology is still a challenge for many vendors, especially when entering the eCommerce space. The more distorted an image becomes compared to the original, the harder it is for a computer to recognize it, as is shown below:
In 2013, the first image-based trademark search engine was released by an Australian company called TrademarkVision.
Table 3. Search & Watch solutions providers embrace image recognition technology
Clarivate took interest, and in 2017, CompuMark released TM go365™ image search, powered by TrademarkVision. TrademarkVision was then officially acquired by Clarivate the next year.
Other companies soon followed suit. It’s harder to precisely date when Corsearch first integrated image recognition into their offerings, but their marketing materials and blog posts suggest that it became a focus in 2020, following the acquisition of Pointer Brand Protection at the beginning of the year. Markify then released their own image-based trademark search in May of 2020. Eleven months later, Corsearch released LogoCheck™, a self-service search tool for screening logos. In January 2022, Fovea IP merged with IPSensus for their image recognition capabilities. At this same time, Huski was successfully completing some final quality assurance benchmarks before going to market. By the official launch in February 2022, Huski’s self-taught model was able to reliably detect over 1,000,000 pictorial brands (logos, design marks) in real-life images, even in very challenging contexts.
5. Self-service and affordability go hand-in-hand
As technology advances enough for better image recognition technology, there’s more opportunity to make trademark clearance and watching more accessible without sacrificing quality. While some providers continue to offer “expert search services,” advancements in AI are enabling attorneys to run high-quality searches on their own.
Table 4. Emergence of self-service trademark clearance and screening tools for attorneys
Even the most skeptical and careful attorneys see the value in self-screening before paying hundreds or thousands of dollars for a full search.
In addition to the established providers entering the self-service search space, many of the new players entering the arena are offering their services at substantially reduced rates thanks to technical advancements. For example, the newest provider, Huski offers unlimited text and image searching for as little as $299 per month. These savings are substantial, considering that a single-class US design search can cost anywhere from $100-$1,000+ with other providers. The promotion of efficient and cost-effective services is an important step forward in improving access to legal services and bettering the IP system for all players.
Disclaimer: This piece is in no way intended to be conclusive, all-encompassing, or the only possible analysis of the industry. It is also not intended to be disparaging of any service providers; we firmly believe that everyone has much to learn from one another about how to make IP better. We seek only to provide a guided introduction to some of the solutions that exist now and the trends that may guide future directions. If you feel as though we have inaccurately represented your company, please contact us and we will happily revise.
About Huski: Huski Inc. is a legal technology startup specializing in solutions that help trademark attorneys as a means of promoting innovation. Its fast-paced team consists of engineers and intellectual property litigation professionals, all of whom strive to increase accessibility to brand protection services for IP lawyers and creators at all stages. Big or small, the brands of an IP firm’s clients deserve protection. Huski believes that AI and big data can upscale IP services efficiently and cost-effectively. For more information and access to a free trial, please visit www.huski.ai.
-
The Shop Safe Act is a bipartisan bill introduced in the House of Representatives on March 2, 2020. The official name of the Act is the Stopping Harmful Offers on Platforms by Screening Against Fakes in E-commerce Act. While the bill has not moved much in Congress yet, rumors are that some in Congress are planning to add this Act to the bipartisan Endless Frontier Act, a technology infrastructure bill already passed in the Senate and could become law in the foreseeable future. If the plan to bundle with The Endless Frontier Act succeeds, it could soon create an earthshaking shock to online marketplaces like Amazon, eBay, or the likes.
What does the Shop Safe Act do? It would amend the Trademark Act of 1946 to give certain e-commerce platforms contributory liability when counterfeits are sold. Liability would apply for counterfeit products that pose a health risk. Under the current law, e-commerce platforms are rarely liable for counterfeit products sold by third parties. However, under the Shop Safe Act, if counterfeit products sold by a third-party seller on the platform pose a health risk to consumers, the trademark owner or customer can hold the e-commerce platform liable for allowing the counterfeit. Given the phenomenal growth in e-commerce during the pandemic and growing online counterfeiting problems, e-commerce platforms could soon be hit with overwhelming counterfeit lawsuits and liabilities if they are not adequately prepared for it.
Liability will be huge for e-commerce platforms under the Shop Safe Act. A court can grant statutory damages between $1000 and $200,000 even in a case without an actual sale of the accused counterfeit products. The damage awards could go up to 2 million dollars if the infringement is found to be willful. Given the significant number of counterfeiting products, damages against e-commerce platforms for non-compliance with the Shop Safe Act could rise astronomically in a short period.
The Shop Safe Act does provide a way out for e-commerce platforms if the third-party seller in question is available to be served a lawsuit in the U.S or the platforms do the following:
(1) required the third-party seller to have a registered agent or a verified address for service of process in the United States;
(2) verified the seller’s identity and contact information;
(3) required as a condition of using the platform that sellers agree not to use counterfeit marks and to consent to the jurisdiction of U.S. courts in any claims related to selling on the platform;
(4) implemented policies to remove and ban repeat offenders; and
(5) implemented technical measures to prescreen listings on the platform and remove listings for goods being sold with a counterfeit mark.
Most popular e-commerce platforms are already doing some or most of the steps (1) through (4). What is lacking is Step №5, i.e., implementing technical measures to prescreen listings on the platform and remove listings for goods being sold with a counterfeit mark. Some platforms like Amazon have already implemented some measures to combat the counterfeiting problem. But that effort is mostly manual and Amazon’s effort to combat counterfeit still largely relies on user or seller reporting/complaints of trademark infringement.
Why have no e-commerce platforms implemented a prescreening procedure for online listings yet? Policing hundreds of millions of listings online is a daunting task, even for a technology powerhouse like Amazon. Not only the volume of listings is too big to manage, but online counterfeiters are also getting “smarter“ and more sophisticated. For example, instead of using famous brand names as part of the listing, counterfeit sellers can use images like the one below to make it difficult for a typical keyword search engine to find a counterfeit product. In this example, the complete phrase “Calvin Klein” is nowhere to be found. Instead, the picture only shows “…vin Klein” and “Calvin….” Counterfeiting is apparent to a person looking at this picture but very difficult for a regular search engine like Google to pinpoint.
Huski.ai, a Silicon Valley-based tech startup, has recently launched an AI-powered branding analysis and monitoring service. Its trademark search engine is potent for complex design mark searches and could become a powerful tool for e-commerce platforms to prescreen carefully disguised online infringers. In the example shown below, the Huski.ai search engine was able to search an online store and accurately locate the infringing symbol on a jersey that counterfeits the logo for the NBA Cavalier team.
Trademark detected on counterfeit goodsThe Huski.ai search engine is also scalable to handle a voluminous number of online listings and stores. With a powerful tool like Huski.ai, e-commerce platforms can rest assured that no “smart” counterfeiter can get away from infringement with canny techniques. Should the Shop Safe Act pass Congress in the near future, e-commerce platforms can quickly become compliant with the law and avoid getting hit by the enormous liabilities if they fail to root out counterfeiters on their platforms.
-
Whether you like it or not, non-fungible tokens (NFTs) have exploded in this universe in the past year. With the advent of Metaverse, the frenzy for NFTs or virtual properties shows no sign of cooling off anytime soon. From art and music to burgers or anything else you can imagine, these digital assets sell like exotic treasures — some for millions of dollars.
However, for the big fashion houses, NFT’s phenomenal growth poses both opportunities and challenges. Famous brands like Luis Vuitton and Burberry have already ventured into the NFT worlds by developing their own digital assets. In the meantime, brand owners are presented with a new and unique challenge in the era of metaverse and NFTs: how to protect their intellectual property rights in the metaverse?
According to The Fashion Law, Hermès recently sent a cease and desist letter to Mason Rothschild, the artist behind the MetaBirkins NFTs. One of the MetaBirkins NFTs is shown below:
NFT created by MetaBirkinsThe creator of the NFT Birkin bags admits his work is spired by Hermès’ famed Birkin bag. Launched by shopping and social platform Basic Space, the NFT Birkin bag sold for $47,000 USD. Following the inaugural Birkin bag NFT, Rothschild created MetaBirkins, a collection of 100 unique Birkin-inspired NFTs. At this time, Rothschild claims that MetaBirkins has sold over 1.1 million USD, which is an amazing feat but not uncommon in the NFT world.
As one of Hermès’ most famous fashion products, a Birkin bag looks like the one shown below.
A Hermès Birkin HandbagThe Birkin bag is so famous, Hermès has spared no expense in protecting its most prestigious product design. It not only registered the Birkin name as a trademark but also managed to register at least two design trademarks in connection with the Birkin bag. One is U.S. Trademark Registration №3,936,105 (“‘105 Trademark”), which covers the contour and overall design of the bag (dotted lines are not part of the trademark). Another is U.S. Trademark Registration №1,806,107 (“‘107 Trademark”), which covers the middle belt with a lock. Both design trademarks are shown as follows:
The benefit of a product design trademark is that it could theoretically protect the design of a famous product forever without worrying about the legal expiration of a design patent and copyrighted work.
The question is: do Rothschild’s MetaBirkin bags infringe Hermès’ trademark rights? The simple answer is that it depends. Although Rothschild publicly admitted that his works were inspired by the Birkin bag, it does not necessarily mean he does not have a good defense to his works. One, he could claim fair use of the Birkin bag trademarks because his designs are artworks, not trademark use. Under the U.S. trademark law, Rothschild could also claim exceptions as artistic works protected under the First Amendment, which he rightfully claimed in an open letter response to Hermès. To claim protection under First Amendment, Rothschild likely has to prove that his use of the Birkin trademarks is (1) “artistically relevant” to the work and (2) not “explicitly misleading” as to the source of the content of the work. Whether this defense works, it will be up to the judge and jury should Hermès decide to take this case to court.
The Hermès v. Rothschild ordeal won’t be the only incident for brand owners. With the flooding of NFTs in the Metaverse, brand owners will have to face the challenge of policing their precious IP against those who seek to profit from it in the virtual world.
The MetaBirkin bag NFTs are relatively easy for Hermès to spot. How about millions of other NFTs currently on sale in OpenSea and other NFT platforms? Do brand owners need a virtual operator to help them scout the virtual world for infringement? Detecting infringement in the virtual poses unique challenges for the tech world. Luckily, we are in the age of artificial intelligence. A.I.-powered tools can be adjusted to meet the new challenges for brand owners in the virtual world. For example, the A.I.-based search engine developed by Huski.ai can be applied to detect “virtual” trademark counterfeiters in the Metaverse. The Huski.ai engine proves to be capable of catching the “smart” counterfeiters on the Internet, who use twisted or cropped images to make it difficult for brand owners to detect infringement. The Huski.ai engine can be redeployed in the virtual world to catch “virtual” infringers using the same technologies. In proving the effectiveness of the search engine, Huski.ai quickly indexed all products listed on Hermes.com, which includes about 10,000 images for a wide range of products, such as handbags, shoes, clothes, accessories, home decors, fragrances, makeups, etc. A couple of search results are displayed below. As the results show, Huski.ai’s engine can effectively find similar Hermès products using the MetaBirkins’ NFT as input. As the NFTs are accepted into the mainstream, “virtual” enforcers will undoubtedly play a more significant role for the big IP houses.
Huski.ai’s search engine identified similar Hermès Birkin products to Metabirkins’ NFTHuski.ai’s search engine identified similar Hermès Birkin products to Metabirkins’ NFT -
Just like taxes, fighting counterfeit is the perennial challenge for brand owners. With online sales taking the dominant role in the 21st century, guarding brand owners’ most precious assets has become more technical than a legal hassle.
According to the U.S. Census Bureau, Total retail sales for the first quarter of 2021 were estimated at $1,581.4 billion, an increase of 7.8 percent from the fourth quarter of 2020. The first quarter of 2021 e-commerce estimate increased 39.1 percent from the first quarter of 2020, while total retail sales increased 16.8 percent in the same period. E-commerce sales in the first quarter of 2021 accounted for 13.6 percent of total sales.
As the dominant online platform for e-commerce, Amazon enjoys continuing fanatic growth to its sales forces. Through its third-party vendor partnership program, hundreds of thousands of new vendors join the platform each year to sell all kinds of products.
Competitors like Shopify and others also enjoy phenomenal growth using their proprietary technologies to help vendors set up online shops. However, for brand owners, the ease of setting up online stores or listings makes the counterfeiting problem far worse than before.
Online counterfeiters are getting quicker and smarter. They are getting more sophisticated with advanced algorithms and targeted marketing techniques. For example, a counterfeit watch seller can quickly set up an online store, send a wave of targeted ads to Facebook users, and reappear under a different name and online store. The swift actions by counterfeiters make it difficult for brand owners to catch up. Instead of using famous brand names as part of the listing, counterfeit sellers can use images like the one below to make it difficult for a keyword searcher to find a fake product. In this example, the complete phrase “Calvin Klein” is nowhere to be found. Instead, the picture only shows “…vin Klein” and “Calvin….” It is apparent to a person looking at this product but challenging for a regular search engine like Google to pinpoint.
The anti-counterfeiting law in the U.S. is already tough enough. Statutory damages could go up to 2 million dollars if the defendant is found willful. Even if no willfulness is found, a court can grant statutory damages between $1000 and $200,000 even in cases without an actual sale of the accused counterfeit products. For brand owners, the difficulty of catching counterfeiters is no longer a legal issue; instead, it hinges on how fast brand owners can locate these online counterfeiters and quickly shut them down through temporary restraining orders issued by the court. This is when AI-powered search engines such as Huski.ai come into play.
Huski.ai, a Silicon Valley-based tech startup, has recently launched an AI-powered branding analysis and monitoring service. Its trademark search engine is particularly powerful for difficult design mark searches and could become a powerful tool for a brand owner to catch carefully disguised online infringers. In the example shown below, the Huski.ai search engine searched an online store and accurately located the infringing symbol on a jersey that counterfeits the logo for the NBA Cleveland Cavaliers team.
Huski also offers intelligent tools for trademark registrability and online infringement analysis. For trademark litigators, Huski’s litigation tool provides valuable insights and docket entries for all trademark infringement cases in U.S. district courts. It could soon relieve a trademark litigator from relying on the clumsy and costly Pacer system.
With a powerful tool like Huski.ai, a brand owner can rest assured that no “smart” counterfeiter can get away from infringement with tricky techniques. If an online marketplace like Amazon or Aliexpress employs an AI-powered tool like Huski.ai to scan its new online listings, can we finally root out the growing problem of online counterfeiting? While it may take a more powerful engine to police an enormous marketplace like Amazon.com, the emergence of Huski.ai and similar tools are indeed off to a good start.
-
The Shortest Intro to Metaverse, Cryptocurrency, NFT, and Web3.0
Metaverse. This term first appeared in the SciFi novel “Snow Crash” written by Neal Stephenson in 1992.
The image is from the Internet, for illustrative purposes only
In the book, metaverse was a virtual urban environment built by top hackers for people to entertain themselves out of dull and busy life in “the future”. When Facebook announced on Oct 28, 2021, that Meta’s new mission is to build this augmented virtual/mix reality world, Metaverse, the company apparently claimed the beginning of making science fiction into reality.
Cryptocurrency is the primary currency used in Metaverse. Just like we use the money to buy goods, we will use cryptocurrency to purchase virtual goods in the metaverse.
The image is from the Internet, for illustrative purposes onlyNFT is the ownership certificate of virtual goods in the metaverse.
People sometimes need ownership certificates to prove they own something in the physical world. A title of a house or a grocery shopping receipt can be an example of such certificates. In the physical world, we don’t always need a receipt after shopping because the goods are physically in our hands after we pay the price. However, in the virtual world, the exchange can only be verified with such certificates, which is why NFT is so important in Metaverse. NFT and Cryptocurrency together will be the foundation of the economy in Metaverse.
Web3.0 is the next version of the Internet where users have clear ownership over their contributions, such as creation, data, and playtime, and get a share of any revenue generated with their contributions.
The image is from the Internet, for illustrative purposes onlyIn contrast, the current Internet, or Web2.0, is a system where user data is maintained by the big companies and for the big companies only. In Web2.0, the primary economy is the targeted ads, whereas the creator economy will be the leading economy in Web3.0.
The Relation among Metaverse, Cryptocurrency, NFT, and Web3.0 is the following. The Metaverse is going to be a big piece of Web3.0. There will be multiple metaverses, such as the numerous websites we use every day, to provide new experiences for users constantly. Creators will build each and every one of the fancy metaverses with virtual worlds, including goods, services, and every detail of them. Such creations will create an economy where consumers use cryptocurrency to do business together and use NFTs to certify the ownership of virtual stuff.
NFTs and Copyright Infringements
As discussed above, the creator economy will play a central role in Web3.0 and Metaverse. Therefore, we need an infrastructure that can empower creators like never before. While other aspects, such as the ease of use of creator tools or the flourishing of the virtual goods marketplaces, have to be prioritized, we see copyright protection as the crown jewel of such creator-friendly infrastructure. No creators will get their fair shares if other people simply download their virtual creations and begin to sell them somewhere else. A healthy ecosystem should, and must, reward creators by either recognization or economic benefit for long-lasting motivation and momentum of growth.
Illustration of Similar Copyrighted Content in ComparisonThe digital assets infringement issue has begun to draw a lot of attention around the globe recently. On Oct 29, 2021, Reuters published an extended discussion of “What are the copyright implications of NFTs?” They cited some copyright laws about ownership relations between creators and owners, and they discussed the responsibilities among sellers and the NFT marketplaces against copyright infringement activities. On Aug 4, 2021, Fortune.com published “Copyright violations could crash the NFT party”. The article discussed the phenomenon of an NFT price dropping dramatically when people found out it was infringing other digital art. On Jun 16, 2021, Techcrunch published “No, NFTs aren’t copyrights”. They had a good and clear discussion about the relation between NFTs (the certificates) and digital assets (the virtual goods). And they clarified that although two parties could use two distinct NFTs to certify two pieces of virtual assets, the assets could have infringement relations to one another. They also said, “Without legal parameters, fraud is inevitable”, referring to such infringements.
Huski Technology and Large Scale Infringement Detection
This previous article explained that Huski.ai exists to detect brand infringements so that brand owners are better informed and brand growth protected. In that article, we also hinted at the vision of everyone being a brand owner in the metaverse. In retrospect, that vision is exactly the creator economy we talked about in this article. Brands, digital assets, and virtual goods are actually interchangeable terms in terms of the entities our technology could process.
Generally speaking, infringement detection or search engine will index an enormous amount of copyright content in multimedia formats and detect their presence in any given query (in the forms of text, image, etc.). Given the nature of the countless copyright content, we can neither build one model for each content nor rely on labeling enough data per content to create AI systems to tackle the problem. It will either end with too many models to handle or too much data to label. Instead, we have to find a scalable way to train as few models as possible to learn the abstract essence of each copyright content about what made them originally unique and try to identify that uniqueness when it appears again in the query. And we have to do it by an unsupervised show and tell. This article has more details about our technology roadmap and progress.
Fortunately, this long-shot technology roadmap paid off. Because we built our technology the way we envisioned, the effectiveness of our search engine is indeed insensible to the content, regardless of representing brands, digital assets, or virtual goods.
That is why Huski is uniquely positioned for Web3.0.