With the launch of ChatGPT (Chat Generative Pre-Trained Transformer), a prototype of chatbot able to converse and write text in natural language developed by the organization OpenAI, the debate on artificial intelligence is livelier than ever.
The context of application of artificial intelligence (A.I) are many. Taking into consideration the context of intellectual property, a question naturally arises: what is the impact of A.I technologies on infringement and protection of the rights of intellectual property?
To answer this question, the European Union Intellectual Property Office (EUIPO) published in march 2022 a study titled “Study On The Impact Of Artificial Intelligence On The Infringement And Enforcement Of Copyright And Designs”, with the aim to evaluate the impact of this new tools both on the application and the infringement of copyright, models and designs.
The methodology used for this analysis is called “Intellectual Property Tech Chain”, according to which the development of an application follows 4 steps (exploration, conversion, weaponization and monetization).
The followed approach can be explained with the “double-edged sword” metaphor: it is based on the consideration that technology can potentially be used both for activity of application of the IP right and for infringement activities, presenting more or less the same characteristics and the same weaknesses, which can then be used by the other and vice versa.
The key points of the study
The study, by analyzing 20 scenarios in order to demonstrate how the A.I. tools can – or could – be used for the application and the infringement of copyright and designs, highlights some key points in terms of:
- Opportunity
- Drivers
- Limitations
- Concerns
A.I technologies can carry out many different functionalities and consequently offer many opportunity of use. Machine learning applied in the content of the protection of IP rights, for example, allows to analyze a great amount of data to intercept threats, scan and identify bots, illicit images and contents, improving the efficiency of the control and enforcement activities.
Natural language processing – another subfield of artificial intelligence, can instead provide and important support in the analysis and block of phishing attacks and in identifying fraudulent behaviors, creating correlation analysis to identify copyright, model and design infringements.
Also computer language and computer vision can turn out to be powerful allies able to identify behavioral patterns and foresee future infringement of intellectual property rights.
Nevertheless, the same tools can be used to carry out design and copyright infringements, to overcome cybersecurity measures and to create illicit or counterfeit products and content.
The study identifies moreover important limitations in the development and use of A.I tools, among which:
- Need for big amount of data: artificial intelligence need a big quantity of data to be ”trained” and work
- Problems in the management of “long-tail problems”:I tools, at the moment, seems to have difficulties in managing “long-tail” problems, as they are dependent on specific scenarios of application.
- Presence of bias: AI tools can carry bias, that is stereotypes or prejudices of the developer who created it, which could therefore provider wrong or, at times, discriminatory evaluations.
- Limited use: even though there has been an important increase of A.I. technologies, their effective use is still very limited in the field of application of copyright, design and models.
Since these technologies can be used as support in taking decisions in very important fields, as in relation to crimes of IP rights infringements, the study highlights that these technologies can potentially have a huge impact on human fundamental rights. Finding a way to use these tools without violating people’s privacy, by defining the legal boundaries for the use of these technologies with the active involvement of the policy makers, becomes extremely necessary.
The investments for the development of artificial intelligences do not seems to slow down. Precisely for this reason, the EUIPO study states that it is possible to foresee an increase of the availability and the use of these tools, both for legal and illicit purposes.
Therefore, increasing the knowledge of these technologies and of their possible uses, for Ip rights enforcement and for infringement activities is extremely important.
Giulia Tibo