The Legal Challenges of Artificial Intelligence

The emergence of artificial intelligence in our daily lives impacts all sectors of activity due to its transversal nature and its anchoring in the digitalization of commerce. The legal understanding of intelligent computer systems is equally delicate due to its recent emergence and singularity.

What are the legal challenges of artificial intelligence?

What is artificial intelligence?

“Artificial intelligence (AI) refers to a set of techniques that allow computer programs to simulate certain traits of human intelligence. The ISO 2382-2015 standard for information technology defines AI as ‘a branch of computer science devoted to developing data processing systems that perform functions normally associated with human intelligence, such as reasoning and learning’.”

AI can take different forms:

• Weak AI refers to intelligent machines reproducing behaviors programmed by humans, which is currently the most widespread form of artificial intelligence. It uses natural language processing like Alexa,

• Machine learning has a learning capacity to optimize its performance,

• Deep Learning goes even further, learning without a prior model,

• Strong AI refers to the ability to reproduce, or even surpass human intelligence in all its components with consciousness and feelings.

AI allows for decision-making and management of certain sectors more efficiently, particularly in marketing, management, customer service… For example, customer relationship management can now be done by an intelligent chatbot.

Its Convergence with Big Data

In recent years, the field of AI has been constantly growing due to technological advances and the phenomenon of big data – the engine of artificial intelligence.

Big data is generally translated as “massive data” or “megadata”. Due to the quantitative explosion of digital data, big data has taken an important place in our society. It involves the search, analysis, sharing, storage, and presentation of data. Big data is based on three vectors called the “3Vs”:

• Volume of considerable data to process,
• Variety of information from various sources,
• Velocity to reach and frequency of generated data.

GenericProductName uses the convergence between artificial intelligence and big data within its software to improve brands’ digital marketing.

What is the Legal Framework for AI?

The legal framework for AI has not yet been defined by legislators through imperative legal norms (hard law).

However, many soft law texts, i.e., non-binding and more incentive legal norms, promote ethical rules inherent to AI. These include charters, codes of ethics, good practice guides, guidelines… The most well-known is the ISO 2382-2015 standard for information technology.

In France, the public debate on digital issues has been opened to formulate proposals for AI regulation. The Law No. 2016-1321 of October 7, 2016 for a Digital Republic entrusted the CNIL with the mission to conduct a reflection on the ethical problems and societal questions raised by the evolution of digital technologies.

The Villani Report of 2018 establishes guidelines for the ethical framework of Artificial Intelligence development. This report recommends, among other things:

• Improving the transparency and auditability of autonomous systems,
• Adapting the protection of rights and freedoms in view of potential abuses related to the use of machine learning,
• Ensuring a system of responsibility in case of damage related to Artificial Intelligence,
• Create a genuine debate forum, diverse and open to society, in order to determine the democratic governance of AI.

This participative body may be enshrined in the “Chamber of the Future”, reforming the Economic, Social and Environmental Council (CESE) to catalyze societal projects such as AI.

Furthermore, Parliament wishes to promote transparency in Artificial Intelligence (Report for a controlled, useful, and demystified Artificial Intelligence by the Parliamentary Office for Scientific and Technological Assessment (OPECST)).

What are the legal protections for AI?

Artificial Intelligence is composed of diverse elements ranging from software and databases to algorithms. Some of these components are protectable by intellectual property.

Copyright

Software and graphical interfaces are generally protected by copyright (Article L. 112-2 of the Intellectual Property Code).

However, algorithms and databases are essential elements for the functioning of these intelligent systems. Therefore, for a company to legally exploit AI in its entirety, it must ensure that the algorithm and databases have been duly acquired, in addition to intellectual property rights.

An algorithm is a set of operational rules whose application allows solving a stated problem through a finite number of operations. The appropriation of algorithms is excluded from copyright:

• “Protecting the functionality of a computer program by copyright would amount to offering the possibility of monopolizing ideas, to the detriment of technical progress and industrial development” (CJEU May 2, 2012, C-406/10);
• The functionalities of software, defined as the implementation of its ability to perform a specific task or obtain a determined result, do not benefit, as such, from copyright protection, as they only correspond to an idea (Cass. civ 1st, December 13, 2005 n°03.21.154);
• “The ideas and principles underlying the various elements of a program, including those underlying its interfaces, are not protected by copyright” (Directive 2009/24/EC of the European Parliament and of the Council of 23 April 2009 on the legal protection of computer programs).

Patents

In principle, the appropriation of algorithms by patent is excluded under Article L. 611-10 of the Intellectual Property Code and Article 52 of the European Patent Convention (EPC).

However, it appears that indirect appropriation of algorithms by patent may be possible when the algorithm is used in a concrete application or included in a full technical process.

Sui generis right for databases

A database can be subject to a proprietary right, provided it constitutes a substantial financial, material, or human investment (Article L. 341-1 of the Intellectual Property Code).

What are the legal obstacles to AI?

Artificial Intelligence has gradually integrated into our daily lives, from optimizing our travel to personalizing our news feeds on social networks or creating intelligent playlists. For example, thanks to AI, the music tracks recorded by the user are analyzed to create a new personalized playlist composed of similar tracks and inspirations. Despite its undeniable advantages, Artificial Intelligence faces certain obstacles such as ethics, compliance with the General Data Protection Regulation (GDPR), security, health, justice…

The regulation of personal data and AI?

AI will naturally increase the domain of personal data exponentially. Therefore, the interests of users must take precedence over the development of AI in order to protect individual freedoms.

What ethics for AI?

The legal news reveals unprecedented challenges between Artificial Intelligence and ethics, particularly concerning abusive language. The law of July 29, 1881, regarding freedom of the press, insults, and defamation, penalizes abusive language provided that a material element, meaning the content of the statements, and an intentional element, meaning the author’s intention to make such statements, can be demonstrated.

Should the operator of intelligent machines be held responsible for abusive language reported through the display of keywords by the latter? This is the question the Court of Cassation has addressed. It firmly establishes the affirmation of the algorithm’s freedom of choice by rejecting the operator’s liability (Cass. civ. 1, June 19, 2013, n°12-17.591).

This issue also raises a legal position on cybersecurity. The latter refers to the ability to protect individuals and the state from any computer threat such as cyberattacks.

What responsibilities for robots?

“The emergence of Artificial Intelligence raises some questions regarding compensation for damages caused by robots. To this end, former Prime Minister Edouard Philippe initiated a national strategy for Artificial Intelligence before the National Assembly in 2017. At the” European level, the Parliament voted on a report on February 16, 2017, establishing the principles of civil liability for damages caused by robots.

In civil matters, determining the person responsible in case of damage caused by a robot is essential. In the current legal framework, a robot cannot be held responsible for acts or omissions when damages are caused to third parties. According to Articles 1245 to 1245-17 of the Civil Code, liability for defective products could apply to damages caused by a robot. However, as soon as Artificial Intelligence would make decisions autonomously, the responsible party could no longer be dissociated, and the defective product liability regime could not apply.

The current legal framework for civil liability of Artificial Intelligence is inadequate due to its autonomy and unpredictability for its operator. The law will need to adapt to technological advances in order to regulate them legally and mutually share risks.

Towards a new legal status: robotic personality?

By analogy with the creation of legal personhood, robot personality could be established with its administrative registration by an identification number to record robots and link them to the responsible natural or legal person. However, the legislator has not yet ruled on the establishment of a specific personality for Artificial Intelligence.

To conclude

Law and Artificial Intelligence are moving at two different speeds. Indeed, the law does “n’t evolve as quickly as AI develops. Today, there are 80 mid-sized companies and SMEs and more than 270 start-ups specializing in AI. France dedicates 400 million euros per year to public funding for AI research. AI is undeniably a future business opportunity for tomorrow’s companies. All that remains is to” wait for the legislator to take a position on the rights and obligations related to AI in order to better understand its legal contours.

Conversely, AI has put itself at the service of law with predictive justice.