[co-author: Cecilia Canova]
As also discussed in our latest bite, we are all eagerly waiting for the imminent official publication of the new draft regulation on AI, which is expected to further define the European Commission approach on the AI technologies.
In order to provide further background and better understand the forthcoming EU Commission draft regulation, we recap below the latest and most significant official positions held by the European Parliament. Indeed, the European Parliament has been extremely productive (and proactive) during the last months: on October 20, 2020 it published the so-called “October Framework on AI”, including among others three reports aimed at addressing how the European Union can regulate AI technologies while promoting innovation, ethical standards and trust, in addition to other resolutions addressing artificial intelligence in criminal matters and in education, culture and the audiovisual sector.
Here is the summary of the three reports.
The European Parliament resolution (and regulation proposal) on “ethical aspects of artificial intelligence, robotics and related technologies” opens the mentioned October Framework on AI.
Considering the potential and the influence that AI, robotics and related technologies have (and may have) within the European Union, both for businesses and citizens, the European Parliament highlights that a human-centric AI should be adopted in order to fully respect human dignity, autonomy and self-determination of the individual. In order to ensure that technologies are used to serve people and not replace or decide for them, as well as to achieve a level-playing field to prevent fragmentation of the internal market, a risk-based approach should be followed, ensuring transparency and accountability.
Moreover, the European Parliament underlines that AI technologies should be designed to eliminate biases and discrimination also among consumers, increase social responsibility and contribute to the achievement of sustainable development (balancing economic, environmental and data-protection perspectives): indeed, a good AI (and data) governance can maintain the balance between safeguarding the rights of citizens and fostering technological development.
Finally, the European Parliament proposes that each Member State should appoint a national supervisory authority responsible for ensuring, assessing and monitoring compliance with the legal obligations and the ethical principles set by the regulations: such authorities will ease the coordination at the Union level, which will ultimately be fundamental to achieve an effective international cooperation also beyond the Union borders.
Who should be deemed liable for AI actions? Which liability rules should be applied? While trying to answer to these questions, in the resolution (and regulation proposal) on the “civil liability regime for artificial intelligence”, the European Parliament calls the European Commission to bridge the existing legal gap when it comes to the liability of the deployers and users of AI-based systems and of AI technologies in general.
Indeed, the complexity, opacity, the capacity of being modified through continuous updates, as well as the capacity for self-learning and the multitude of actors involved constitutes a striking challenge to the effectiveness of any European Union and national liability regime.
Furthermore, considering the liability of the operator (be it a backend operator or a frontend operator), the European Parliament believes that that liability rules should cover all operations of AI-systems, regardless of where the operation takes place and of its physical or virtual nature. The type of AI-system remains a key factor: following the distinction already proposed by the European Commission, the European Parliament provides for a strict liability regime for the so-called high-risk AI systems, recommending the European Commission to exhaustively list and regularly update all high-risk AI-systems. On the other hand, operators of non high-risk systems should be subject to a fault-based liability regime.
Finally, the European Parliament considers liability insurance as one of the key factors that defines the success of new technologies, products and services and highlights that publicly funded compensation mechanisms are not an appropriate answer to the rise of AI.
The third parliamentary resolution completing the October Framework on AI provides guidance on “intellectual property rights for the development of artificial intelligence technologies”.
The European Parliament highlights that patent law should be reviewed / re-examined in light of the development of AI (see here our bite on AI and IP protection): indeed, even though algorithms, mathematical methods and computer programs are not patentable as such, they may form part of a technical invention that can be patented and all the economic operators, among which European start-ups, should be aware of this opportunity.
Besides patent law, all intellectual property rights should be assessed: by way of example, with regards to copyright law, the notion of originality could represent an obstacle to the protection of AI-generated creations and the notion of “creative work” should be reconsidered.
Lastly, the European Parliament considers essential to boost the sharing of data generated in the European Union in order to stimulate innovations in the AI field.