by Valérie Simonart
Partner (Brussels Bar, Liedekerke)
Professor (Université Libre de Bruxelles)
Table of contents
THE 2017 PROPOSAL FOR A DIRECTIVE ON CIVIL LAW RULES ON ROBOTICS CALLING
FOR A SPECIFIC LEGAL STATUS FOR ROBOTS
IMPORTANCE OF THE QUESTION
THE GRANTING OF LEGAL PERSONALITY TO AI: A SPECIFIC QUESTION AMONG MANY
THE LEGAL POSSIBILITY TO GRANT LEGAL PERSONALITY TO AI
THE DESIRABILITY TO GRANT LEGAL PERSONALITY TO AI
1. THE 2017 PROPOSAL FOR A DIRECTIVE ON CIVIL LAW RULES ON ROBOTICS CALLING FOR A SPECIFIC LEGAL STATUS FOR ROBOTS
On 16 February 2017, the European Parliament enacted a resolution requesting the Commission to submit a proposal for a directive on civil law rules on robotics1. The Parliament called on the Commission to explore all possible legal solutions, such as “creating a specific legal status for robots in the long run, so that at least the most sophisticated autonomous robots could be established as having the status of electronic persons responsible for making good any damage they may cause, and possibly applying electronic personality to cases where robots make autonomous decisions or otherwise interact with third parties independently” (§ 59, f)
The very idea of granting legal personality to robots has caused an uproar.
Hundreds of robotics experts, industry leaders, law, medical and ethics experts have signed an “Open letter to the European Commission. Artificial intelligence and robotics”, denouncing that recommendation as relying on “a robot perception distorted by Science-Fiction and a few recent sensational press announcements” and being inappropriate “from an ethical and legal perspective”2.
This resolution was short-lived as the European Parliament has abandoned the very idea of granting legal personality to artificial intelligence (hereinafter, “AI”). In 2019, it adopted a resolution on a comprehensive industrial policy on AI and robotics3, which does not consider anymore any special status for AI and focusses on a piecemeal approach in the areas of law impacted by AI. In 2020, it adopted a resolution with recommendations to the Commission on a civil liability regime for artificial intelligence, in which it expressly states that “Any required changes in the existing legal framework should start with the clarification that AI-systems have neither legal personality nor human conscience” and that “all physical or virtual activities, devices or processes that are driven by AI-systems may technically be the direct or indirect cause of harm or damage, yet are nearly always the result of someone building, deploying or interfering with the systems; notes in this respect that it is not necessary to give legal personality to AI-systems”4.
2. IMPORTANCE OF THE QUESTION
There is a dramatic gap between AI and the law: while AI is developing very fast, the pace of law-making is terribly slow.
The slow pace of law-making is not necessarily a bad thing: “il ne faut toucher aux lois que d’une main tremblante” (Montesquieu); the enactment of laws relies on a democratic process which involves proposals, explanations, discussions, objections, amendments, rebuttals, votes. For example, the recent reform of the Code on companies in Belgium, which has been characterized – and criticized! – by the speed of the legal process, has taken five years between its announcement and the vote of the law.
Meanwhile, technology is unfolding exponentially (the so-called “Intelligence Explosion”) to the point that, according to some scientists, it might become uncontrollable (the so-called “Technological Singularity”).
Of course, as rightfully put by the authors of the open letter to the Commission, one has to guard against misconceptions arising from science fiction but, at the same time, one may not ignore this singularity and has to prepare for the advent – even if not certain – of AI on a level approaching human intelligence (or even exceeding it).
3. THE GRANTING OF LEGAL PERSONALITY TO AI: A SPECIFIC QUESTION AMONG MANY
Artificial intelligence raises a lot of exciting questions in all kinds of fields and in many areas of law, be it certification standards, products liability, personal data privacy, consumer protection, intellectual property rights, etc.
In a contribution in the book Entre tradition et pragmatisme. Liber amicorum Paul-Alain Foriers, published by Larcier in April 2021, I have examined the very specific question of the granting of legal personality to AI.
As mentioned above, the European Parliament has abandoned the very idea of granting legal personality to AI. Anyway, as repeatedly held by the European Court of Justice, “companies are creatures of national law” and a directive could not compel the States to grant AI legal personality.
Even in Estonia, one of the boldest Member States in terms of technology, a proposal to adopt a special AI law aimed at granting legal personality to AI has been rejected.
However, for a very long time, legal personality has been granted to ships in the United Kingdom and, in other parts of the world, to idols, rivers or even the entire ecosystem.
More recently, Saudi citizenship has been granted to a robot, named Sofia5.
In the aforesaid contribution in the Liber amicorum Paul Alain-Foriers, I explain that it has been denounced as a marketing stunt and that citizenship has been granted to Sofia in infringement of the laws of Saudi Arabia. I also give similar examples, such as a robot running for mayor ... and securing thousands of votes in the mayoral election.
I also explain that the debate on the very possibility to grant legal personality to AI is often clouded by a confusion between what is possible and what is desirable, an anthropomorphic reasoning based on free will or even metaphysical considerations about the existence of a conscience, which is reminiscent of the discussions on the legal personality of companies in the nineteenth century.
4. THE LEGAL POSSIBILITY TO GRANT LEGAL PERSONALITY TO AI
From a theoretical point of view (as opposed to what is desirable in practice), nothing stands in the way of granting legal personality to AI. Legal personality is only a concept that designates the ability of an entity to have rights and obligations. It does not imply that a legal person is akin to a human being, that it has a free will or a conscience or that it must enjoy all the rights granted to human beings. It is no more unthinkable to grant legal personality to AI than to a company.
All that it would take, at least in the countries (such as Belgium) that apply a legalistic system according to which only the law determines which entities are legal persons, is a law stating that provided that certain conditions are met, AI may acquire legal personality. Of course, such a law would raise a lot of issues, such as, for instance the starting and ending dates of AI’s legal personality, the determination of AI’s assets, the possible liability of other persons for AI. They are not insurmountable: in the past, they have also arisen for companies and they have been solved.
5. THE DESIRABILITY TO GRANT LEGAL PERSONALITY TO AI
A completely different question, which can become very passionate and emotional, is whether it is desirable to grant legal personality to AI.
Many are opposed, on the ground that that AI lacks the component of soul, will or awareness, that it would be contrary to human dignity to grant legal personality to AI but not to foetuses, that it raises the risk of moral hazard and opportunities for abuse or still on other grounds discussed in my aforesaid contribution.
However, here again, one must keep in mind that legal personality is a technical concept and not a human attribute and that its granting does not depend of any value judgment but merely on the assumption that it will promote in general the common well-being and, in particular, the goals pursued by the legislator.
In addition, the introduction of a special status for AI would not per se exonerate the makers from their liability. But, admittedly, faced with the difficulty of identifying the guilty party, it could be tempting to shift the blame on the AI alone.
In my aforesaid contribution, I question the opportunity of granting legal personality to AI in its present state of development, while the dangers are all too apparent.
Indeed, although the development of AI impacts many fields of law, the issue of the recognition of the legal personality mostly arises in discussions regarding liability as the self-learning abilities of AI make it hard to apply product liability rules because of the difficulty to draw the line between the inherent defects of a product and the damages caused by AI in the process of developing that product, thereby making the traditional rules insufficient. The problem extends to criminal liability: there already have been several accidents involving AI and causing injury and even death of human beings.
On 21 April 2021, the European Commission has adopted a proposal for a Regulation laying down harmonised rules on artificial intelligence (Artificial Intelligence Act)6, which is part of a wider comprehensive package of measures that address problems posed by the development and use of AI, including liability issues7 and which makes no mention whatsoever of the legal personality of robots.
Indeed, one wonders how the legal personality of the AI would help the victim. As long as the AI does not have assets, it cannot subscribe an insurance (who would pay the premiums) and a judgment rendered against the AI would be useless; to make a claim against a guarantee fund, it is not required that the AI be a legal person.
And if AI would be able to acquire and accumulate assets, whatever the short-term benefits of granting legal personality to AI and allowing them to own assets, they would rapidly be outweighed by the danger of AI controlling the economy.
While it would be possible to grant legal personality to AI, it would not solve any issues raised by their development and would, most probably, create new problems, ranging from the risk of shifting liabilities to AI, to the danger of misuse of AI or even, in a far-reaching scenario, of AI misusing its capabilities.
AI will probably become an unavoidable tool for most aspects of our life, but the mere fact that robots can be built to look like human beings and, in certain domains and to a certain extent, have better cognitive capacities than human beings, is no reason to grant them legal personality.