A New Player in Liability Law: Artificial Intelligence

Erdem & Erdem Consultancy Ltd | View firm profile


The use of ​​artificial intelligence is increasing day by day and this technology is starting to play an active role in our lives. In addition to the use of artificial intelligence in common products such as smartphones, 3-D printers or drones, there are advanced technology breakthroughs that demand our attention. Unmanned aerial vehicles, self-driving cars, or robots performing surgery, are examples which reveal what this technology is capable to do. Although the technological developments from the Industrial Revolution to the age of Industry 4.0 are quite exciting, the damages that artificial intelligence can cause and who will be affected are still unclear. This newsletter article examines the current developments in Turkish and foreign law regarding liability for damages caused by artificial intelligence.

Definition and Legal Status of Artificial Intelligence

There are many different definitions of artificial intelligence. Known for its active work in the field of artificial intelligence, the EU Commission defines artificial intelligence as “systems that have a certain autonomy and exhibit intelligent behavior by analyzing and acting on their environment to achieve certain goals.”[1] John McCarthy, who is known as the founder of artificial intelligence, defines artificial intelligence as “the science and engineering of making intelligent machines.”[2] There is no unified regulation on the definition of artificial intelligence yet. However, it is useful to touch on the concepts of “machine learning” and “deep learning” to better understand what artificial intelligence means. Machine learning, a term often used as synonymous with artificial intelligence, is actually a field of artificial intelligence that uses algorithms to learn from data. These algorithms build a model based on the inputs and use the resulting insights to make decisions or make predictions. Deep learning, on the other hand, is a much more recent concept and is defined as “continuous learning”. It exemplifies the methods used by the human brain to solve complex problems.”[3]

While it is quite difficult to define artificial intelligence, there is also no consensus on its legal status. Indeed, there is no specific legislation or definition regarding artificial intelligence in Turkish law yet. For this reason, it should be evaluated whether artificial intelligence falls within the scope of the definitions already in the legislation. Although there are various opinions which define artificial intelligence as a person or a slave, there are important criticisms in the doctrine against these opinions, and there are ethical considerations as well.[4] Since goods must have a tangible existence in accordance with Turkish law, it is also not widely accepted that artificial intelligence technology falls under the definition of goods.

On the other hand, there are opinions which classify artificial intelligence as a product or service.[5] The definition to be made here varies according to the nature of artificial intelligence and the services it offers, since the development level and capacity of each artificial intelligence is different. While it is possible to consider artificial intelligences with more advanced and complex actions as services, those with simpler and plainer features can be described as “products”. On the other hand, there are opinions that artificial intelligence can be considered as a commodity within the scope of the Consumer Protection Law No. 6502 (“CPLN”).[6] Pursuant to Article 3/h of the CPLN (Definitions) goods are defined as movable property, immovable property for residence or holiday purposes, and software, audio, video and similar intangible goods prepared for use in electronic environment, which are subject to trade.

Among the foreign opinions on artificial intelligence, the Civil Law Rules on Robotics published by the European Parliament Legal Affairs Committee in 2017 draws attention. According to this report, due to the unique nature of artificial intelligence, it is recommended to create a new legal status and create an “electronic personality”.[7] Thus, it is understood that the European Parliament is closer to the definition of “person” rather than the definitions of products, goods or services, and wants to create a sui generis personality.

Liability for Damages Caused by Artificial Intelligence

Contractual relations and torts, which form the fundamental basis of legal responsibility, are evaluated in the doctrine in terms of the damages potentially caused by artificial intelligence. Firstly, considering contractual liability, the responsibility arising from the contract between the buyer and the seller of the artificial intelligence shall be examined. In contractual liability, there are various advantages, such as providing ease of proof to the injured party and the statute of limitation is 10 years. However, in accordance with the principle of privity of contract, it will not be possible for third parties other than the buyer to make a claim against the seller. As a matter of fact, the defects that occur in artificial intelligence will most likely be caused by the manufacturer’s fault rather than the seller; however, it will not be possible to sue the manufacturer due to the principle of privity.[8]

As another source of responsibility, tortious liability and strict liability are frequently evaluated in the doctrine. Pursuant to Article 49 of the Turkish Code of Obligations No. 6098 (“TCO”), a person who causes harm to another by a wrongful and unlawful act is obliged to compensate for this damage. Compensation may be claimed under the provisions of the tortious liability, in the event that one of the persons who had no previous relationship with the tort causes harm to the other.[9] For this reason, if a person who is harmed due to artificial intelligence does not become a party to any contractual relationship and the conditions required by law are fulfilled, compensation can be claimed according to this general provision.[10]

Strict liability cases are a type of liability regulated exceptionally in our legislation. In this type of liability, a person is held liable only for the damages caused by his own actions, or by some related facts, without any showing of fault.[11] Categories of strict liability under Turkish law include equity liability, the duty of care, and liability for dangerous activities. Among these, there are several opinions which argue that artificial intelligence should be accepted as a personality in order to apply the equity liability.[12] Liability for dangerous activities is regulated under Article 71 of the TCO (Liability for dangerous activities and equalization). Accordingly, if a loss arises from the activity of a business that poses a significant danger, the owner of the business and the operator, if any, are jointly responsible for this loss. In the event that artificial intelligence is used in an enterprise where the danger is envisaged in this article, it is possible to bring an action for danger liability.

Finally, the topic of producer responsibility should be addressed. Today, it is generally accepted that the manufacturer may be held responsible for damages caused by artificial intelligence. Since manufacturers have the technical knowledge and expertise required for the production of artificial intelligence, it is considered quite reasonable to hold them responsible[13]. As stated above, while it is not possible for the buyer to sue the manufacturer due to the privity of the contract, the manufacturer’s responsibility provides a legal basis for that. In this regard, the responsibility of the manufacturer is clearly regulated in accordance with the Product Safety and Technical Regulations Law No. 7223. Accordingly, the responsibility of the manufacturer comes to the fore in cases where a product put on the market is faulty and this causes damages. However, some authors are of the opinion that the causal link may pose a problem since causality must be proved by the injured party[14]. Given the complexity and technical side of artificial intelligence, buyers are likely to have difficulty proving this causation.

Developments in Foreign Legislation

It is well known that the EU Commission is closely following the artificial intelligence issue and is working to create a uniform regulation in this field. In this context, in the White Paper[15] published in 2020, the artificial intelligence ecosystem was closely examined and opinions were expressed about the legislation to be created and what the responsibility regime should look like. It is noteworthy that if a very strict liability regime regarding artificial intelligence liability is developed, the development of this technology may be hindered. In this context, the EU Commission prepared the first draft legislation on artificial intelligence in 2021.[16] In the aforementioned draft, a “risk-based” approach has been adopted and levels such as unacceptable risk, high risk, low risk and minimum risk have been determined according to the risk level of artificial intelligence activities. The proposal is expected to enter into force in 2023. Thus, an important step will be taken towards a uniform legislation specific to artificial intelligence.


Artificial intelligence, the use of which is rapidly increasing along with its capabilities, is generally considered too complicated to be defined with current legal concepts. As there is no uniform legislation on this subject in our country and in the world yet, attempts are being made to solve the problems related to artificial intelligence with existing concepts and responsibility regimes. There are various debates about how artificial intelligence should be regulated through legislation. A general caveat, however, is that the development of this technology could be hampered if overly burdensome liability regimes are introduced. It is eagerly awaited that the uniform regulations on how to define artificial intelligence. And how it will be subject to rules, will be clarified in the near future.

(Authored by Yagmur Zeytinkaya and first published by Erdem & Erdem on August 2021)

[1] https://digital-strategy.ec.europa.eu/en/library/communication-artificial-intelligence-europe (Access date: 17.07.2021)

[2]  http://jmc.stanford.edu/artificial-intelligence/what-is-ai/ (Access date: 09.08.2021)

[3]  Büyüközkan Feyzioğlu, Gülçin: Gelişen Teknolojiler ve Hukuk II: Yapay Zeka, 2021, s. 6-7

[4]  Bak, Başak; Medeni Hukuk Açısından Yapay Zekanın Hukuki Statüsü ve Yapay Zeka Kullanımından Doğan Hukuki Sorumluluk, 2018, s. 9

[5]  Sarı, Onur: Yapay Zekanın Sebep Olduğu Zararlardan Doğan Sorumluluk, TBB Barosu 2020 Sayı: 147, 2019, s. 259

[6]  Sarı, s. 262

[7]  European Parliament resolution of 16 February 2017 with recommendations to the Commission on Civil Law Rules on Robotics (2015/2103(INaL)) https://www.europarl.europa.eu/doceo/document/TA-8-2017-0051_EN.html#title1 (Access date: 18.07.2021)

[8]  Sarı, s. 265

[9]  Tandoğan, Haluk: Türk Mesuliyet Hukuku, 2010, s. 6

[10]  Sarı, s. 265

[11]  Tandoğan, s. 8

[12]  Sarı, s. 268

[13] Sarı, s. 275

[14] Kapancı, s. 177

[15]https://ec.europa.eu/info/sites/default/files/commission-white-paper-artificial-intelligence-feb2020_en.pdf  (Access date: 20.07.2021)

More from Erdem & Erdem Consultancy Ltd