On October 20, 2020, the Members of the European Parliament have adopted three resolutions providing recommendations to the European Commission aiming to guide it in the drafting of future regulations on Artificial Intelligence (AI). These guidelines are specifically aimed at fostering innovation while preserving ethics and confidence in new technologies (1), the protection of intellectual property rights (2) and the responsibility of the various stakeholders (3).

  These three initiatives come in the wake of the Commission's publication on February 19, 2020 of a White Paper on AI and a report on the safety and liability framework for AI and the circulation on July 17, 2020 of the results of the public consultation that took place between February and June 2020 on the issue of AI. They foreshadow the legislative proposals that the Commission is expected to present on the subject in early 2021.    

1. The ethical framework ON artificial intelligence

The first resolution adopted in response of the report presented by MP Ibán García del Blanco includes recommendations and a proposal for a regulation on ethical principles in the development, deployment and use of AI, robotics and related technologies. More specifically, it is based on the following principles and objectives:

• To build confidence in AI among all actors, especially when they are considered high risk ;
• To promote the development, deployment and use of AI within the Union, in particular by providing developers and users with an appropriate regulatory framework that promotes legal certainty while guaranteeing fundamental rights, consumer protection and respect for ethical principles;
• To promote transparency and improve the exchange of information between citizens and within organizations, developing, deploying or using AI to ensure that it complies with Union law, its values and fundamental rights and ethical principles.

On the basis of these principles, the proposal for a Regulation presents 24 articles divided between ethical principles, obligations relating to high-risk technologies and control by European and national bodies, and advocates in particular:

• Human-centred AI, developed and controlled by humans (Articles 5 and 7) ;
• Control of high-risk AI, in particular by :
  • mandatory assessment of its conformity (Article 6),
  • guarantees relating to its security, transparency, traceability and finally the responsibility of its developers, deployers and users (Article 8),
  • guarantees against bias and discrimination on grounds such as race, gender, sexual orientation or religion (Article 9),
  • social responsibility and gender equality so that AI does not, in particular, contribute to the dissemination of misinformation (Article 10),
  • environmentally sustainable AI (Article 11) ;
  • respect of privacy and limitation of the use of technologies for remote identification in public places, such as biometric or facial recognition (Article 12);
  • a right to compensation for persons who have suffered damage caused by their development, deployment or use (Article 13),
  • a methodology for the assessment of high-risk technologies and the establishment of an assessment of their compliance with the obligations set out in Articles 6 to 12 (Articles 14 and 15);
• Establishment of an European ethical compliance certificate (Article 16);
• Establishment in each EU Member State of an independent public body to monitor the application of this Regulation (Article 18).

 

2. Protecting intellectual property rights

The second resolution, adopted as a result of a report presented by MP Stéphane Séjourné, consists of 22 recommendations to preserve the EU's digital and industrial sovereignty, ensure its competitiveness and protect innovation and intellectual property rights. It states that it would be necessary, in particular, for these purposes :

• To understand the various dimensions of AI through definitions both technologically neutral and flexible enough to apply to tomorrow's technological developments and future uses;
• An effective intellectual property regime for the digital age at European level, including strong safeguards to protect the EU patent system against illegal activities that harm innovative developers;
• Focus on a sectoral and type-specific assessment of the intellectual property rights implications of AI technologies, taking into account in particular the degree of human intervention, the autonomy of AI, the importance of the role and source of data and copyright protected content;
• Not to endow AI technologies with legal personality[1];
• To make a distinction between human creations assisted by AI, for which the current copyright framework remains applicable, and autonomous creations generated by AI which must be protected under the legal framework of intellectual property rights in order to encourage investment but which may not benefit from copyright protection, in order to respect the principle of originality linked to a natural person, except that rights are granted to natural or legal persons who created the work[2];
• Promote more transparent and open access to certain non-personal data and databases in the EU and data interoperability, in particular by creating a Single European Data Space;
• Establish design systems involving processes of control, review, transparency, accountability and verification of AI decision-making processes with human intervention;
• Enhance media awareness and education, as well as the availability of AI technologies to verify information and fight against hyper-destruction, in particular by facilitating the tracing of the use of copyrighted works.
• Strengthen media awareness and education, as well as the availability of AI technologies to verify information and combat hyper-destruction by facilitating, in particular, the tracing of the use of copyrighted works.

 

3. A special liability regime in case of damage caused by the iA

The third and final resolution adopted following a report presented by MP Alex Voss includes recommendations and a proposed regulation on a civil liability regime for the operation of AI systems. In particular, it is specifically grounded on the following principles and objectives:

• Achieving a balance between protecting citizens and supporting investment in innovation;
• Ensuring maximum legal certainty throughout the chain of liability, including for the producer, the operator, the injured parties and any other third party;
• Encourage the introduction of new concepts, rather than replacing existing liability regimes;
• Preserve confidence in new technologies by offering citizens the same protection regime whether the harm is caused by an AI system or not, whether it is of a physical/material or virtual/immaterial nature.

The proposed Regulation sets out 14 articles applicable "where a physical or virtual activity, device or process controlled by an AI system has caused harm or damage in the form of injury to the life, health or physical integrity of a natural person or to the property of a natural or legal person, or has caused significant intangible damage resulting in verifiable economic loss". It provides more specifically :

• Concerning high-risk AI systems :
  • A regime of strict liability of their operators for all damage caused by such AI systems, unless they can prove that they acted with due diligence or that the damage was caused by an autonomous activity, device or process controlled by such AI systems and except in cases of force majeure (Article 4);
  • The obligation for front-end and back-end operators to ensure that the operation of AI systems and their services are covered by liability insurance (Article 4);
  • Maximum amounts of compensation according to the type of damage caused and the methods of calculating compensation (Articles 5 and 6);
  • Specific limitation periods (Article 7).

• Concerning the other AI systems: a system of liability for the fault of their operators, with the exception of proving that the damage was caused through no fault of their own and except in cases of force majeure (article 8);

• Rules for sharing liability in the event of concurrent or exclusive fault of the injured party (Article 10) and in the event of a multiplicity of operators (Article 11), whose shares of liability will be determined in proportion to the degree of control exercised by each one over the risk linked to the operation and functioning of the AI system (Article 12).

 

The three resolutions therefore call on the Commission to establish a harmonised regulatory framework at EU level in the field of AI by adopting rules that are flexible enough to keep pace with the rapid development of new technologies. In particular, they call for the right balance to be struck between protecting citizens and supporting investment and the development of innovation by giving priority to innovative, human and European AI.

    [interne id="66984"][interne id="67036"]   [1] On this point, our contribution against the creation of a legal personality for AI (Article Editions FrancisLefevere of 18 April 2018) [2] On this point, your contribution on page 101 of the CSPLA report on the legal and economic stakes of artificial intelligence in the cultural creation sector (Report of 27 January 2020 under the direction of Professors AlexandraBensamounand JoelleFarchy