Understanding Artificial Intelligence and Enforcement of Rights
With John Zacharia
In the June, 2021 edition of the BPP, I explored the European Union Intellectual Property Office (EUIPO) Study of the International Judicial Cooperation in Intellectual Property Cases report with leading authority John Zacharia who was an integral part of the study team. (See A Deep Dive into International Cooperation Tools for Brand Protection)
In 2022, John continued his work with the EUIPO as a member of United Nations Interregional Crime and Justice Research Institute (UNICRI) team that authored another key study:
Impact AI of the Infringement and Enforcement on Copyright and Designs
Having received a high volume of “reads” of that previous interview, we decided to explore this new study through a similar Q & A format – but please be sure to click on the report in its entirety as it is a comprehensive exploration of this important subject matter.
Welcome AGAIN, John!
As with the previous report, this new EUIPO Study is full of valuable information for the brand protection community. Specifically, it examines how we enforce IP rights in this age of new, emerging, and disruptive technologies, specifically discussing artificial intelligence or AI.
I think the first aspect I want to call out to the readers is the Definitions Section. It may seem strange to do so as definitions may be perceived as dry and a bit uninteresting, but new technologies bring new terms we need to understand. As such, we are all now dealing with new unfamiliar terms and acronyms. So I urge the readers to spend some time in that section or be sure to refer to it as you read through the content of the Study.
So, John, let’s once again do our best to break down important aspects of this Study for brand protection practitioners, starting with why was it important for the EUIPO to explore the impact of artificial Intelligence on IP in particular?
I think the first aspect I want to call out to the readers is the Definitions Section. It may seem strange to do so as definitions may be perceived as dry and a bit uninteresting, but new technologies bring new terms we need to understand. As such, we are all now dealing with new unfamiliar terms and acronyms. So I urge the readers to spend some time in that section or be sure to refer to it as you read through the content of the Study.
So, John, let’s once again do our best to break down important aspects of this Study for brand protection practitioners, starting with why was it important for the EUIPO to explore the impact of artificial Intelligence on IP in particular?
Thanks, Leah, and it’s always a pleasure to chat with you! I was honored to join another UNICRI team in producing and authoring this Study on AI for EUIPO. As you know, EUIPO does a great job of taking both a visionary and a practical approach to IP – looking ahead at how emerging issues may impact intellectual property enforcement and infringement today and in the near future.
Although the concept of “AI” goes back to at least the 1950s, AI has evolved so rapidly that it has started to affect our everyday lives. Not surprisingly, AI’s rapid evolution impacts IP, too, and AI’s impact on IP will only increase in the near future. So, it’s to EUIPO’s credit that they commissioned a report analyzing AI’s current and future impact on IP at exactly the time when these issues are moving to the forefront for IP owners.
Although the concept of “AI” goes back to at least the 1950s, AI has evolved so rapidly that it has started to affect our everyday lives. Not surprisingly, AI’s rapid evolution impacts IP, too, and AI’s impact on IP will only increase in the near future. So, it’s to EUIPO’s credit that they commissioned a report analyzing AI’s current and future impact on IP at exactly the time when these issues are moving to the forefront for IP owners.
This Study starts with a discussion and representation of the “double-edged” sword of IP opportunity and IP threat with four steps of methodology in between. What should practitioners consider as instructive in this metaphor?
Like most emerging technologies, AI technology by itself is neutral in the sense that AI is neither inherently a threat to, nor inherently an opportunity to protect, IP rights. What drives the impact of a particular form of AI technology is how it is designed, and how it is used.
At its essence, AI technology could be characterized as a “prediction machine.” How useful and how effective that prediction machine will be depends principally on two things: the biases of the designer of the particular AI technology being assessed and the quality of the data fed into that prediction machine (i.e., the better the data, the better the predictions). Once a particular form of AI technology has been assessed and enabled by the data the AI technology receives, it is available for use. At this point, it is entirely up to the user (or in this metaphor, the “sword wielder”) how that AI technology will be used. Hence, the double-edged sword metaphor: AI can be designed and weaponized by criminals to facilitate IP infringement, but it can also be designed and utilized by law enforcement and IP owners to detect and to prevent IP violations.
At its essence, AI technology could be characterized as a “prediction machine.” How useful and how effective that prediction machine will be depends principally on two things: the biases of the designer of the particular AI technology being assessed and the quality of the data fed into that prediction machine (i.e., the better the data, the better the predictions). Once a particular form of AI technology has been assessed and enabled by the data the AI technology receives, it is available for use. At this point, it is entirely up to the user (or in this metaphor, the “sword wielder”) how that AI technology will be used. Hence, the double-edged sword metaphor: AI can be designed and weaponized by criminals to facilitate IP infringement, but it can also be designed and utilized by law enforcement and IP owners to detect and to prevent IP violations.
The Study primarily addresses threats to copyright and designs – are there any aspects to the Study that can be applied to trademark infringement and counterfeiting? Specifically, do any of the 20 different scenarios from the two storylines used to illustrate infringements, prove instructive for TMs? If so, which ones?
I’m glad you brought up the wide variety of fictional scenarios in the Study. To me, what most distinguishes this Study from others is the fact that it is not limited to describing the different types of AI technology that could impact IP. This Study goes beyond that sometimes highly technical discussion by using 20 fictional (yet realistic) scenarios to illustrate how different types of AI technology could potentially be used by IP criminals to infringe IP and by law enforcement to detect and protect against such infringement. In this way, the scenarios can offer lessons to all IP owners – including brand owners – about the threats and opportunities that AI poses to their IP rights.
As you indicated, the scenarios are divided into two storylines: one involving physical products and one involving digital content. In my view, most of the scenarios in the storyline focused on physical products (“Storyline α“ in the Study) would be both familiar and instructive to brand owners. For example, Storyline α’s various mass production, importation, physical marketing, and online marketing scenarios could easily be replaced by counterfeit or trademark infringing goods so that they become fictional scenarios involving “mass production of counterfeit goods,” “importation of counterfeit goods,” “physical marketing of counterfeit goods,” and “online marketing of counterfeit goods.” These are very familiar scenarios to brand owners and law enforcement alike. Indeed, some of these scenarios arguably arise in the real world more frequently in the context of trademark counterfeiting than in the copyright or design contexts. Most important, the discussion of how AI-enabled technology could be used to facilitate copyright, or design, infringement in these scenarios will generally apply equally to how this same technology could be used to facilitate trademark counterfeiting. The same would generally be true of these scenarios’ discussion of how law enforcement, or trademark owners, could use AI-enabled technology to detect and to prevent attempts to counterfeit trademarks.
As you indicated, the scenarios are divided into two storylines: one involving physical products and one involving digital content. In my view, most of the scenarios in the storyline focused on physical products (“Storyline α“ in the Study) would be both familiar and instructive to brand owners. For example, Storyline α’s various mass production, importation, physical marketing, and online marketing scenarios could easily be replaced by counterfeit or trademark infringing goods so that they become fictional scenarios involving “mass production of counterfeit goods,” “importation of counterfeit goods,” “physical marketing of counterfeit goods,” and “online marketing of counterfeit goods.” These are very familiar scenarios to brand owners and law enforcement alike. Indeed, some of these scenarios arguably arise in the real world more frequently in the context of trademark counterfeiting than in the copyright or design contexts. Most important, the discussion of how AI-enabled technology could be used to facilitate copyright, or design, infringement in these scenarios will generally apply equally to how this same technology could be used to facilitate trademark counterfeiting. The same would generally be true of these scenarios’ discussion of how law enforcement, or trademark owners, could use AI-enabled technology to detect and to prevent attempts to counterfeit trademarks.
As with other new emerging technologies, there are aspects to be monitored for facilitation of infringement. Which ones should BP practitioners be most aware of?
For this Study, we interviewed more than 37 experts on AI and on how AI impacts IP to help us better understand how new AI-enabled technologies are impacting the infringement and enforcement of IP today, and how they are most likely to impact IP in the near future. The Study highlights the types of AI-enabled technologies that these experts emphasized could have the greatest impact on the infringement and enforcement of IP. For example, the Study’s scenarios highlight how IP criminals could use AI-enabled machine learning, computer vision, and/or natural language processing to scan websites to identify the most popular designs and select which designs to infringe. Trademark counterfeiters could use the same AI-enabled technologies to identify and target which trademarked goods that they want to counterfeit and then traffic in those counterfeit goods. Trademark counterfeiters could even use AI-enabled virtual reality and augmented reality to develop a realistic online persona (similar to what is described in the scenarios) to promote counterfeit products on social media marketplaces while at the same hiding their identities.
Another broader point for brand owners to consider is that organized crime groups no longer need a group member who is a technology or cyber expert to take advantage of some of these AI-enabled tools. Today, cybercriminals have refined their methods to the point where they can make their work available to others through so-called “Cybercrime-as-a-Service” or “Crime-as-a-Service” (CaaS). This evolution of CaaS means that IP criminals can outsource the use of AI-enabled tools to facilitate their counterfeiting or infringing activity. In the course of preparing the Study, we also learned that some cybercriminals even take an “off-the-shelf” approach by offering all-in-one online toolkits that IP criminals who are less technologically savvy can purchase and use on their own to facilitate their counterfeiting activities.
Another broader point for brand owners to consider is that organized crime groups no longer need a group member who is a technology or cyber expert to take advantage of some of these AI-enabled tools. Today, cybercriminals have refined their methods to the point where they can make their work available to others through so-called “Cybercrime-as-a-Service” or “Crime-as-a-Service” (CaaS). This evolution of CaaS means that IP criminals can outsource the use of AI-enabled tools to facilitate their counterfeiting or infringing activity. In the course of preparing the Study, we also learned that some cybercriminals even take an “off-the-shelf” approach by offering all-in-one online toolkits that IP criminals who are less technologically savvy can purchase and use on their own to facilitate their counterfeiting activities.
There are also aspects of AI that can be used in developing effective detection and/or enforcement strategy. The Study seems to outline many applications of AI? What are most relevant to brand protection?
The Study emphasizes that AI is not just a tool for IP violators. The Study’s scenarios illustrate how law enforcement and brand owners could use any number of AI-enabled tools to facilitate detection of IP violations and enforcement of IP rights, too. For example, brand owners could use various types of AI-enabled machine learning to analyze large amounts of data, especially online, to identify threats, scan images to identify websites or images trafficking in counterfeit goods, and identify patterns of counterfeiting activity. More specifically, brand owners could use AI-enabled deep learning and convolutional neural networks to analyze and recognize logos and other relevant images and then use that data to scan shapes and patterns in online listings for similarities with registered trademarks. Brand owners could use AI-enabled tools like these to better identify and analyze an image of a good being offered for sale online that bears a counterfeit mark identical to a registered trademark – even if the IP criminal deliberately placed the image at an angle to avoid discovery by less advanced online tools used by brand owners.
AI-supported, blockchain technology is another AI application that the Study’s scenarios show could be used by IP owners. Brand owners could use such technology to secure a label, code, or image that would enable customers to verify the authenticity of a product. Brand owners could even choose to share this information with law enforcement to aid a criminal IP investigation.
AI-supported, blockchain technology is another AI application that the Study’s scenarios show could be used by IP owners. Brand owners could use such technology to secure a label, code, or image that would enable customers to verify the authenticity of a product. Brand owners could even choose to share this information with law enforcement to aid a criminal IP investigation.
What key findings from this Study are most important for brand protection practitioners to understand?
First and foremost, I think it’s important for brand owners to understand that, like any other new high-impact technology, AI is a neutral tool. In terms of the “threat” side of the double-edged sword, I hope the Study helps brand owners and other IP owners understand better which AI-enabled technologies counterfeiters and other IP criminals could employ to commit IP crimes and other IP violations.
Although the “threat” side of the double-edged AI sword tends to get the most attention, brand owners should not lose sight of the fact that they too could utilize AI to stop counterfeiters. I think the Study’s fictional scenarios crystallize how law enforcement and IP owners could use AI-enabled technologies. For instance, brand owners can use AI to identify and prioritize risks and even guide incident responses by using machine learning to analyze large amounts of data to detect counterfeiting threats. Brand owners can use the same AI technology to more efficiently scan images to identify sites selling counterfeit goods and detect patterns of counterfeit goods sales.
Although the “threat” side of the double-edged AI sword tends to get the most attention, brand owners should not lose sight of the fact that they too could utilize AI to stop counterfeiters. I think the Study’s fictional scenarios crystallize how law enforcement and IP owners could use AI-enabled technologies. For instance, brand owners can use AI to identify and prioritize risks and even guide incident responses by using machine learning to analyze large amounts of data to detect counterfeiting threats. Brand owners can use the same AI technology to more efficiently scan images to identify sites selling counterfeit goods and detect patterns of counterfeit goods sales.
Lastly, what did you find most interesting in this Study? Maybe something that you learned or that you had not thought about before being involved in this Study?
The most interesting thing I learned is not only the sheer scope of AI-enabled tools that could be used to infringe and enforce IP, but also how much cheaper and easier to use these AI-enabled tools have become. With so many new AI-enabled technologies emerging every year, it can be hard to keep up with the changes. In working on the Study, it became clear that we needed the 20 scenarios to highlight the ways in which AI-enabled technologies could be used to facilitate the infringement and enforcement of IP in ways that we would not have expected just a few years ago. I hope the Study helps brand owners and IP owners get a head start on, and a better understanding of, the increasing impact AI-enabled technologies will have on the infringement and protection of IP.
THE BRAND PROTECTION PROFESSIONAL | JUNE 2022 | VOLUME 7 NUMBER 2
2022 COPYRIGHT MICHIGAN STATE UNIVERSITY BOARD OF TRUSTEES
- Next…