The aim of these articles is to keep our Clients and Associates updated about developments in the sector of Intellectual Property in general and our firm in particular. In this way, we wish to provide a broader view of the tools that the field of trade marks, domain names, patents, designs and related rights offers to entrepreneurs to enhance and protect their efforts in researching and developing new solutions and ideas.
Artificial Intelligence: The Impact of the AI Act on AI Patents
The Regulation (EU) 2024/1689 issued by the European Union and effective 1 August 2024 (so-called "AI Act") is the first regulation in the world that sets rules to artificial intelligence, a technology which is increasingly impacting on the lives of us all.
Artificial intelligence has powerfully emerged as a tool to simplify the creation of content - text, images, video, audio, software code, etc. - in response to specific user requests. While this is already enough to raise the question of the authorship of the content produced by artificial intelligence systems, as well as concerns about the legitimate use of the enormous amount of data (often subject to copyright) used to instruct such systems, even greater risks arise from the potential use of this technology to spread disinformation, increase public scrutiny, discriminate, commit crimes and fraud, and ultimately even endanger human life itself.
Against this background, the Regulation aims "to improve the functioning of the internal market” and "to promote the uptake of human centric and trustworthy artificial intelligence (AI) while ensuring a high level of protection of health, safety, fundamental rights as enshrined in the Charter of Fundamental Rights of the European Union (the ‘Charter’), including democracy, the rule of law and environmental protection, to protect against the harmful effects of AI systems in the Union, and to support innovation."
In this Regulation, the European Union has set out a number of requirements that AI systems must meet depending on the level of risk involved in their use: the higher the risk of infringing individual and collective rights, the more stringent the requirements.
In short, there are four levels of risk covered:
- Unacceptable risk: contradiction of EU values and principles, such as respect for human dignity, democracy and the rule of law. AI systems that present this kind of risk are forbidden to those who are not involved in higher national interests, such as military and national security;
- High risk: significant impact on people's fundamental rights or security. AI systems with this level of risk include those used for personnel selection and recruitment, admission to education, and the provision of essential social services;
- Limited risk: possibility of influencing the rights or wishes of users, but to a lesser extent than high-risk systems. AI systems with this level of risk include those used to generate or manipulate audiovisual content (such as deepfakes), or to provide personalised suggestions (such as chatbots);
- Minimal or no risk: no direct impact on people's fundamental rights or security, and the provision of wide margins of choice and control to users. AI systems presenting this level of risk include those used for recreational and aesthetic purposes, such as video games or filters for editing photographs.
The requirements applicable to high-risk AI systems include to establish, implement and document a risk-management system, i.e. a continuous iterative process planned and executed throughout the life cycle of an AI system and aimed to eliminate or reduce the risks associated with the use of AI systems.
Among other requirements for high-risk AI systems, it is worth stressing the obligation to provide a detailed description of the elements of the AI system and the process related to its development, including the training data sets used (e.g. information on their origin and how they were obtained and selected). Still speaking of high-risk AI systems, there is also an obligation to register them in a special database, draw up a declaration of conformity and obtain the EC marking.
For AI systems with low or minimal risk, there is instead the obligation to inform users that they are talking to a chatbot and not a human, and the obligation for images and texts to contain the information that they have been generated by an AI system.
The Regulation also introduces fines for violations of the established obligations.