AI Liability in Russia: Experimental Legal Regimes and Autonomous Systems
Russian law on artificial intelligence liability is developing through a combination of experimental legal regimes (eksperimentalnyye pravovyye rezhimy), Civil Code amendments, and sector-specific regulation. The Russian approach to AI liability reflects a cautious but pragmatic approach, prioritising the development of AI technologies within controlled regulatory frameworks while gradually adapting general principles of civil liability to the challenges posed by autonomous systems.
Experimental Legal Regimes
Federal Law No. 258-FZ of 31 July 2020 on Experimental Legal Regimes in the Sphere of Digital Innovation provides the framework for testing AI technologies in controlled regulatory environments. An experimental legal regime (eksperimentalnyy pravovoy rezhim) allows for the temporary suspension or modification of certain regulatory requirements for participants in the experiment, enabling the testing of AI technologies that may not fully comply with existing legal rules. The experimental regime is established by a decision of the Government of the Russian Federation for a specified period (not exceeding three years, extendable by up to one year). Participants in the experimental regime must comply with requirements established by the programme of the experimental regime, which may include mandatory insurance, information disclosure, reporting obligations, and measures to protect the rights of third parties. The experimental regime for autonomous vehicles has been the most significant application of this framework, with designated testing zones in Moscow, the Republic of Tatarstan, and other regions.
Civil Code Amendments for Autonomous Systems
The Civil Code of the Russian Federation (Grazhdansky Kodeks, GK RF) has been amended to accommodate AI systems, though there is no comprehensive regulation of AI liability. The concept of a digital person or electronic person has not been adopted in Russian law. Instead, existing civil law concepts of legal personality, fault, causation, and damages are applied to AI-related harm, adapting traditional principles to the specific characteristics of autonomous systems. The Civil Code provisions on tort liability (Chapter 59) apply to harm caused by AI systems, with the operator or owner of the AI system generally bearing liability. The concept of source of increased danger (istochnik povyshennoy opasnosti) under Article 1079 of the Civil Code — which imposes strict liability on the owner of a source of increased danger for harm caused by that source — has been applied to autonomous vehicles and other AI systems, treating them as inherently dangerous activities requiring heightened liability. The Russian courts have not yet developed a substantial body of case law specifically addressing AI liability, and the application of general tort principles to AI systems remains an area of legal uncertainty.
Insurance Requirements
Insurance plays a central role in the Russian approach to AI liability. The experimental legal regimes regulations typically require participants to obtain liability insurance covering potential harm caused by AI systems. For autonomous vehicles operating under experimental regimes, mandatory insurance requirements have been established, with minimum coverage amounts determined by the Government. The insurance framework aims to ensure that victims of AI-related harm receive compensation while allowing AI development to proceed within manageable risk parameters. The insurance industry in Russia has begun developing specialised insurance products for AI-related risks, including autonomous vehicle liability insurance and professional liability insurance for AI system developers. The regulatory framework for AI insurance remains in development, with proposals for mandatory insurance requirements for high-risk AI systems being discussed as part of broader AI regulation reform.
The Moscow AI Experiment
The Moscow Artificial Intelligence Experiment (Moskovsky eksperiment po iskusstvennomu intellektu), launched in 2019, has been the most significant practical test of AI regulation in Russia. The experiment, conducted under the framework of the experimental legal regime, authorised the testing of AI technologies in various public service domains, including healthcare diagnostics, medical image analysis, transport management, and urban planning. The Moscow experiment generated valuable data on the performance, reliability, and risks of AI systems in real-world applications. The results informed the development of regulatory requirements for AI systems, including standards for data quality, algorithmic transparency, human oversight, and liability allocation. The experiment demonstrated the potential of AI technologies to improve public service delivery while also highlighting the need for robust regulatory frameworks to address risks of error, bias, and harm. The lessons from the Moscow experiment have influenced the development of federal AI regulation and have been referenced in international discussions of AI governance.
Autonomous Vehicle Regulation
Autonomous vehicle regulation represents the most advanced area of AI-specific legislation in Russia. Federal Law No. 443-FZ of 29 December 2022 on the Organisation of the Operation of Highly Automated Vehicles (HAVs) established the legal framework for the operation of autonomous vehicles on public roads. The law defines highly automated vehicles as vehicles equipped with an automated driving system that performs the dynamic driving task without the need for human intervention. The law establishes requirements for the certification of autonomous vehicles, the licensing of operators, the registration of HAVs, and the liability of owners and operators. Under the law, the owner of an autonomous vehicle bears civil liability for harm caused by the vehicle, regardless of whether the harm resulted from a system malfunction, an unexpected environmental condition, or a design defect. Manufacturers and software developers may be held liable under product liability principles if the harm results from a defect in the vehicle or its software. The law also establishes mandatory insurance requirements and creates a compensation fund for victims of autonomous vehicle accidents.
Data Protection and AI
AI regulation in Russia intersects with data protection law, particularly the data localisation requirements of Federal Law No. 242-FZ. The processing of personal data for AI training and operation must comply with the requirements of the Federal Law on Personal Data, including the requirement to store personal data on servers located in Russia. The Federal Law on Information, Information Technologies, and Information Protection establishes requirements for the processing of information, including the use of AI for automated decision-making. The development of AI systems for facial recognition, predictive policing, and social scoring has raised privacy concerns, and the regulatory framework for such applications remains under development. The Moscow facial recognition system — deployed on a large scale since 2020 — operates under specific legal authorisation but has raised questions about the adequacy of legal protections for privacy and data protection.
Significance
The Russian approach to AI liability is characterised by a preference for experimental legal regimes, the adaptation of existing civil law principles, and the use of insurance mechanisms to manage risk. The absence of comprehensive AI-specific liability legislation reflects both the early stage of AI development in Russia and the preference for regulatory flexibility through experimental frameworks. As AI technologies become more widely deployed, the development of more comprehensive liability rules will be necessary. The emerging regulatory framework balances the promotion of AI innovation with the need to protect individuals’ rights and provide compensation for AI-related harm. The experience of the Moscow AI Experiment and the autonomous vehicle regulation will inform the development of future AI regulation, including potential legislative proposals for a comprehensive AI law.