By akademiotoelektronik, 28/08/2022

Artificial intelligence and contract: towards an upheaval of vices of consent?

It was therefore in the face of a legal and ethical vacuum that the President of the European Commission Ursula von der Leyen affirmed that she wanted "that the application of these new technologies be worthy of the trust of our citizens. We encourage a responsible, human-centric approach to AI”[1]. It is with this ambition of transparency that a proposal for a European regulation was presented on April 21, 2021 setting, for developers, specifications to be complied with in terms of AI in order to guarantee a framework and legal security for European users[2]. Partly targeted as a high-risk AI system, the use of AI in contract management raises questions.

The integration of AI into contractual processes now exists in all stages of the life of the contract. Thus, AI can be used from assistance in drafting the contract to contractual monitoring tasks, including termination issues. However, the level of complexity of implementing AI in contract management tools varies depending on the substance that the AI ​​addresses. A distinction can therefore be made between AI systems for assistance with drafting and contract monitoring, on the one hand, and autonomous contract management systems, on the other.

Integration of AI and validity of the contract

The validity of a contract meets, under the terms of the new article 1128 of the Civil Code, the following three conditions: (1) free and informed consent of the parties, (2) their capacity to contract and (3) lawful and certain content. What then of the applicability of these conditions, and in particular of the consent of the parties, when artificial intelligence revolutionizes contractual practices?

The issue of defects in consent, within the meaning of Article 1130 of the Civil Code, thus seems essential in light of the use of AI. However, its study should be restricted to certain scenarios. Take, for example, the AI ​​systems used in contract drafting, which in fact make it possible to assist the drafter of a deed in identifying the elements to be inserted in the contract and transcribing them using a standard clause. It cannot be excluded the hypothesis of an imperfect transcription of the information communicated to the AI ​​so that the contract could be tainted with error. Who would be responsible for such an error?

Let's focus first on the hypothesis of the use of an AI system in the drafting of a contract. As a reminder, the drafting of contracts, not in authentic form, drawn up for others on a regular and remunerated basis, and creating rights or obligations, falls within the framework of the monopoly of lawyers[3]. Thus, the use by the lawyer of an AI system in the drafting of a contract must comply with the duty to advise and the precautionary principle which are imposed on it. It follows that the production of a contract tainted with an error could thus engage the lawyer's liability insofar as negligence was committed. It would be open to the lawyer in such a case to bring a recourse action against the publisher of the AI ​​system, at the risk of being confronted with watertight disclaimers. In the same way, when the parties are the sole writers of the act, the interplay of the non-responsibility clauses of the publishers of AI systems imply a responsibility of the users - in other words the parties themselves - in the verification of the information contained in the document thus produced.

It remains to be seen the consequences of the development of autonomous AI systems allowing the creation of a contract between a user and the AI ​​directly.

A new legal tool: the “smart contract”

The traditional view of the contract, defined by article 1101 of the Civil Code as an agreement of will, between two or more people, intended to produce legal effects (legal regime established by the reform resulting from ordinance no. 2016-131 of February 10, 2016 reforming contract law, the general regime and proof of obligations), however, sees its effectiveness questioned by the arrival of new contracts that integrate artificial intelligence in their formation and execution: smart contracts.

A smart contract is a computer program that runs in the blockchain system[4]. These programs are created by developers in the form of conditions. These conditions are then checked by the program. If they are respected, they produce effects pre-established by the computer code. Thus the smart contract runs automatically as soon as the previously defined conditions are met.

To secure the contract, this information is entered in the blockchain; the data is therefore impossible to modify or delete. The blockchain is a digital data storage and transmission system that works through the internet. This system is secure because each user has a public key and a private key to access it. The public key is known to everyone and allows a transaction to be carried out. The private key is used to decrypt a transaction, hence the need to keep it secret. To date, smart contracts are largely generated on the Ethereum Blockchain.

Smart contracts are currently used in particular in the fields of insurance, medical data protection, financial services and real estate. The user may therefore find himself concluding a contract through a simple exchange with the AI ​​system, most often taking the form of a digital dialogue (“chatbot”).

In the field of insurance, smart contracts can be used for the purpose of automatic compensation of victims of damage when the conditions are met. For example, and although the platform was discontinued in 2019, the insurance company Axa launched the Fizzy platform in 2017, which automatically compensates an insured in the event of cancellation or flight delay. .

In the field of real estate, the use of smart contracts could extend to seasonal rentals. It could, for example, be a smart rental contract by which the tenant could book, read the conditions of the lease, and pay automatically as soon as he enters the premises.

The smart contract is therefore an asset and a major development thanks to the automation it puts in place, by reducing, for example, the risk of non-payment at the time of the execution of the contract. It also promotes a certain technical security through the use of blockchain technology which anchors the terms of the contract as soon as it is formed, without it being possible to delete or modify them afterwards. Finally, smart contracts eliminate the intermediaries and trusted third parties, such as lawyers, usually necessary to draft the contract.

The smart contract: a tool compatible with the law of obligations?

Do these new contracts provide more legal certainty? Is common contract law adapted to these new smart contracts?

Currently, two doctrinal currents oppose each other. One is firmly convinced that the common law of contracts resulting from the 2016 reform is a sufficient framework for the formation and execution of smart contracts thus subject to the provisions of articles 1101 to 1239 of the Civil Code[5].

The other, however, is more mixed[6]. Indeed, smart contracts can be the source of legal uncertainty, especially at the time of their formation. Due to automaticity, the condition set by article 1128 of the Civil Code relating to the consent of the parties could be called into question. First, it would be necessary to be able to ensure that the consent exists as soon as the smart contract is formed. In practice, this remains difficult since the developers are the only ones capable of coding the conditions of the contract. Then, it would be necessary to be able to ensure the integrity of the consent of the parties, that is to say that it is not vitiated. In French law, the error on the essential qualities of the service or of the co-contractor is likely to vitiate the consent and thus leads to the nullity of the contract (article 1132 of the Civil Code). However, in the case of a smart contract, the original will of the contracting party may not be reflected. The essential quality of the service, for example when it comes to a product or a service, may be different from that expected, and this leads to legal uncertainty when this essential quality was decisive for the consent.

It follows that the asymmetry of information emerges as a decisive problem of smart contracts with regard to the uncertainties weighing on the obligations incumbent on AI, in particular in terms of the pre-contractual obligation of information. Indeed, this obligation introduced in article 1112-1 of the Civil Code forces the party holding information that it knows to be decisive for the consent of the other party, to communicate to it whether this ignorance was legitimate or if this party confidence in its co-contractor. The breach of such an obligation may lead to the consideration of fraud against the uninformed party, the legal uncertainty weighing on both parties is significant.

However, the fraud hypothesis in smart contracts is difficult to undertake. Indeed, fraud, within the meaning of article 1137 of the Civil Code, requires demonstrating the will to deceive of the person to whom it is opposed. However, the demonstration of a fraudulent intention emanating from the AI ​​is not easy in that it would imply proving that the initial coding of the system and its learning components were intended to knowingly mislead a contracting party. In other words, how to determine the existence of a fraudulent intent through a line of code? There is no doubt that such a demonstration will be difficult to implement, especially since it could then be challenged by the designer of the AI ​​upon presentation of all the factual elements that led the system to conclude these conditions the contract with the user. It would therefore be a question for the party deeming himself injured to prove the existence of fraud, even though his opponent will be able to justify in extenso his intentions during the formation of the contract by simple transcription of the data collected by the AI.

It is undeniable that artificial intelligence will be increasingly integrated into the legal tools used by both litigants and practitioners, thus auguring an upheaval in the legal environment of the contract, from its formation to its execution. This digital revolution, like all the others, will certainly make certain tools and concepts obsolete. At a time when defects in consent are increasingly rarely put forward as effective legal grounds before the courts, the challenges raised by artificial intelligence seem to encourage this trend. It remains to be seen whether the interference of artificial intelligence in the contractual field will sound the death knell of the current vision of the defects of consent.

Chronicle “Augmented Law, Jurist and Practice of Law”

The objective of this column is to deal with current issues relating to this transformation. In a context where digital, big data and data analytics, machine learning and artificial intelligence are profoundly and lastingly transforming the practice of law, creating "augmented lawyers" but also calling for "augmented law" with regard to challenges and new business models driven by digital.

With its Augmented Law Institute, EDHEC Business School has a major advantage in positioning the knowledge, skills and function of the lawyer at the center of transformations in business and society. It is defined around 3 strategic areas of development: its offer of hybrid training, its research useful to the legal industry, its Legal Talent Management platform. https://www.edhec.edu/fr/ledhec-augmented-law-institute

---

[1] “The EU for “responsible” and human-controlled artificial intelligence”, L’Express, February 19, 2020.

[2] Proposal for a Regulation of the European Parliament and of the Council laying down harmonized rules on artificial intelligence (Artificial Intelligence Law) and amending certain Union legislation, https://eur-lex. europa.eu/resource.html?uri=cellar:e0649735-a372-11eb-9585-01aa75ed71a1.0020.02/DOC_1&format=PDF.

[3] Article 54 of the law of December 31, 1971, https://www.legifrance.gouv.fr/loda/article_lc/LEGIARTI000039280601/

[4] A blockchain is a database that contains the history of all exchanges between its users since its creation. This database is secure and distributed: it is shared by its various users, without intermediaries, which allows everyone to check the validity of the chain

[5] E. Viguier, Blockchain and smart contracts: technological, legal and business issues: LexisNexis, CDE 2017, interview 2

[6] Bruno ANCEL, Smart contracts: societal revolution or new Pandora's box? LexisNexis, Electronic Commerce Communication n° 7-8, July 2018, study 13.