Biotechnology Therapeutic Ai Oversight In Russia.
1. Introduction
Therapeutic Artificial Intelligence (AI) oversight in Russia concerns the legal regulation, approval, deployment, monitoring and accountability of AI systems used for medical diagnosis, treatment recommendations, drug development, patient monitoring and therapeutic decision-making.
Therapeutic AI is increasingly relevant to biotechnology because it can analyse genomic information, identify disease patterns, predict treatment responses, assist in pharmaceutical research and support personalised medicine.
In Russia, therapeutic AI oversight is not governed by one comprehensive AI-specific medical statute. Instead, it operates through a combination of legislation dealing with:
Healthcare and medical activities.
Medical-device registration and safety.
Personal data and medical confidentiality.
Clinical trials and pharmaceutical regulation.
Patient rights and informed consent.
Civil liability and professional responsibility.
Experimental legal regimes for digital innovation.
The central legal question is:
Who is legally responsible when an AI system recommends or contributes to a medical treatment that causes injury to a patient?
The answer depends on the AI system's function, regulatory classification, the conduct of the healthcare provider, the manufacturer's obligations and the circumstances of the patient's treatment.
2. Legal Framework Governing Therapeutic AI in Russia
A. Federal Law No. 323-FZ of 21 November 2011 – Fundamentals of Healthcare Protection
This is the principal legislation governing healthcare activities in Russia.
It establishes the legal framework for:
Medical treatment and healthcare services.
Patient rights and medical confidentiality.
Informed voluntary consent.
Medical intervention and professional responsibilities.
Quality and safety of medical care.
Regulation of medical devices.
Article 19 recognises the patient's right to medical care and information about their health.
Article 20 governs informed voluntary consent to medical intervention. This is particularly important when AI systems influence treatment recommendations.
Article 38 deals with medical devices. Software intended for medical diagnosis, treatment or monitoring may qualify as a medical device and require state registration.
Legal implication: A hospital cannot treat an AI recommendation as an automatic substitute for legally required medical assessment and patient consent.
B. Government Resolution No. 1684 of 30 November 2024 – Medical Device Registration
This resolution establishes the modern framework for state registration of medical devices.
It came into force on 1 March 2025 and provides for:
Technical testing.
Safety and effectiveness assessment.
Clinical investigations where required.
Examination of manufacturer evidence.
Registration-dossier requirements.
Procedures for changes to registered medical-device software.
Importantly, the registration rules expressly address software using artificial intelligence that qualifies as a medical device.
Changes to AI medical software can therefore have regulatory consequences. A manufacturer cannot assume that every software update is merely an ordinary technical modification.
imdrf.org+2
C. Federal Law No. 152-FZ of 27 July 2006 – Personal Data Protection
Therapeutic AI systems often process:
Patient medical histories.
Genetic information.
Diagnostic images.
Laboratory results.
Biometric information.
Treatment-response information.
Articles 5, 6, 9 and 10 are particularly relevant.
Article 5 establishes principles concerning the purpose, proportionality and retention of personal-data processing.
Article 6 establishes the legal grounds for processing.
Article 9 governs consent.
Article 10 regulates special categories of personal data, including health information.
Legal implication: A biotechnology company developing an AI-based cancer-treatment recommendation system must establish an appropriate legal basis for collecting and processing patient information. It must also protect that information against unauthorised access and use.
static-0.minzdrav.gov.ru+1
D. Federal Law No. 61-FZ of 12 April 2010 – Circulation of Medicines
This legislation is relevant where therapeutic AI is used in:
Drug discovery.
Pharmaceutical development.
Clinical investigations.
Medicine registration.
Pharmaceutical safety monitoring.
Evaluation of medicinal products.
An AI system may help identify a potential drug candidate or predict drug interactions. However, an algorithmic prediction does not itself establish that a medicine is legally authorised, clinically effective or safe for human use.
E. Federal Law No. 86-FZ of 5 July 1996 – State Regulation in the Field of Genetic Engineering
This legislation is relevant to biotechnology applications involving genetic engineering and genetically modified biological materials.
Where therapeutic AI is integrated into genetic engineering research, developers must consider the applicable biosafety, research and genetic-engineering requirements.
The legal classification of the underlying biotechnology activity remains important even when AI performs the computational analysis.
3. Role of Russian Regulatory Authorities
Росздравнадзор высылает предостережение за ФРМО и ФРМР
Roszdravnadzor – Federal Service for Surveillance in Healthcare
Responsible for medical-device registration, regulatory supervision, safety monitoring and oversight of medical-device circulation.
ВСЕРОССИЙСКИЙ ТЕРАПЕВТИЧЕСКИЙ КОНГРЕСС С МЕЖДУНАРОДНЫМ УЧАСТИЕМ "БОТКИНСКИЕ ЧТЕНИЯ"
Ministry of Health of Russia
Develops healthcare policy, medical standards and regulatory requirements concerning medical practice and healthcare technologies.
Roskomnadzor
Supervises compliance with personal-data and information-protection legislation.
Federal Antimonopoly Service
May become relevant where AI healthcare procurement, market access, competition or discriminatory commercial practices raise antitrust issues.
4. Registration and Approval of Therapeutic AI Systems
The regulatory classification of a therapeutic AI product is one of the most important legal questions.
Not every AI application used in a hospital is automatically a medical device.
| AI application | Potential regulatory treatment |
|---|---|
| Hospital appointment chatbot | Generally administrative software |
| Medical-record summarisation | Depends on intended purpose and functionality |
| AI cancer-detection software | Potential medical device |
| AI treatment recommendation system | Potential medical device |
| AI radiology-analysis software | Potential medical device |
| AI drug-discovery platform | Depends on its intended function |
| AI patient-monitoring software | Potential medical device |
The manufacturer's intended purpose, actual functionality and medical claims are important in determining classification.
Registration process
Define the intended medical purpose.
Determine the applicable medical-device classification.
Conduct technical testing.
Establish safety and performance evidence.
Complete clinical investigations where required.
Submit the registration dossier.
Obtain the necessary state registration.
Maintain post-market safety monitoring.
Under Article 38 of Law No. 323-FZ, medical devices generally must be registered before being placed into circulation in Russia, subject to applicable statutory exceptions.
roszdravnadzor.gov.ru+2
5. Special Oversight Requirements Introduced in 2026
Important regulatory development
One of the most significant developments in Russian therapeutic AI regulation is Roszdravnadzor Order No. 123 of 10 February 2026.
It establishes a procedure for automatically transmitting information about AI-based medical software to Roszdravnadzor's Automated Information System (AIS).
The Order:
Was registered with the Ministry of Justice on 8 May 2026.
Was officially published on 12 May 2026.
Entered into force on 23 May 2026.
Remains in force until 31 December 2027.
The requirements apply to AI software classified as a medical device where it has the relevant built-in automatic data-transfer function.
The information includes:
Medical-device name and registration details.
Software version.
Medical application and type of solution.
Number of examinations or objects processed.
Results generated by the AI system.
Number and nature of system errors.
Identification of the medical organisation using the software.
The principal responsibility for ensuring transmission rests with the manufacturer or its authorised representative.
This development is particularly important because it introduces a more structured mechanism for regulatory observation of AI medical-device performance after registration.
roszdravnadzor.gov.ru+3
Example:
Suppose a Russian hospital uses AI software to analyse 20,000 radiological examinations.
If the software repeatedly fails to identify early-stage cancer, the operational information and error records can become important for regulatory investigation, safety assessment and corrective action.
This illustrates the importance of post-market monitoring rather than relying exclusively on pre-market approval.
6. Therapeutic AI and Patient Rights
Patient protection is a fundamental part of therapeutic AI oversight.
A. Right to informed consent
Under Article 20 of Federal Law No. 323-FZ, informed voluntary consent is generally required before medical intervention.
Where AI substantially influences a treatment decision, healthcare organisations should provide understandable information about:
The proposed medical intervention.
Its expected benefits and risks.
Available treatment alternatives.
The role of AI in supporting the decision.
The involvement of the treating medical professional.
The precise extent of AI-specific disclosure is not comprehensively established by a separate Russian statutory provision.
B. Right to medical confidentiality
Article 13 of Law No. 323-FZ protects medical confidentiality.
AI developers and healthcare organisations must ensure that access to patient records, genetic information and diagnostic results is appropriately controlled.
C. Right to quality medical care
Article 10 of Law No. 323-FZ establishes the principle of accessibility and quality of medical care.
A hospital should not justify an inadequate treatment decision simply by claiming that its AI software generated the recommendation.
D. Right to challenge medical decisions
Patients may use applicable healthcare complaint procedures, seek independent medical assessments and pursue judicial remedies where their rights have been violated.
7. Liability for Therapeutic AI Errors in Russia
Russian law does not generally recognise an AI system as an independent legal person capable of bearing criminal responsibility.
Liability must therefore be assessed according to the conduct of the human and legal persons involved.
| Responsible party | Potential legal responsibility |
|---|---|
| Physician | Negligent clinical decisions and failure to exercise appropriate professional judgment |
| Hospital | Defective medical services and organisational failures |
| AI manufacturer | Defective or unsafe medical software |
| Software developer | Liability where its legally attributable conduct causes harm |
| Medical-device distributor | Relevant product circulation and safety obligations |
| Data-processing organisation | Personal-data protection violations |
A. Civil liability
Relevant provisions of the Russian Civil Code include:
Article 1064 – General grounds for liability for causing harm.
Article 1068 – Liability of an employer for harm caused by employees.
Articles 1095–1098 – Liability for harm caused by defects in goods, work or services.
Articles 151 and 1099–1101 – Compensation for moral harm.
For example, if defective AI software produces a dangerous chemotherapy recommendation and the defect causes patient injury, civil liability may arise depending on the defect, causation, fault and applicable statutory conditions.
B. Criminal liability
The Russian Criminal Code may become relevant in serious cases.
| Provision | Potential relevance |
|---|---|
| Article 109(2) | Causing death through negligent improper performance of professional duties |
| Article 118(2) | Negligently causing serious bodily injury through improper performance of professional duties |
| Article 238 | Providing services that do not meet legally applicable safety requirements, where the statutory elements are established |
| Article 293 | Official negligence by an appropriate public official |
Criminal liability cannot be imposed merely because an AI recommendation was incorrect. Investigators must establish the required elements of an offence, including the relevant person's conduct, fault and causal connection with the harm.
8. Important Russian Case Laws and Judicial Decisions
Judicial precedents and legal analogies
An important qualification is necessary: there is not yet a well-established body of published Russian judicial decisions directly determining liability for injuries caused by therapeutic AI or AI-driven biotechnology treatment.
Nevertheless, existing medical-negligence and patient-protection decisions provide important principles for future disputes.
Case 1. Supreme Court of the Russian Federation – 10 April 2019
Case No. 11-UD19-1
Subject: Negligent performance of professional medical duties.
The Supreme Court examined a criminal conviction involving a medical professional accused of causing death through improper performance of professional duties.
The Court emphasised that a finding of criminal responsibility requires identification of the specific professional requirements violated and examination of whether those violations caused the patient's death.
Legal significance:
A criminal conviction cannot rest merely on the fact that treatment was unsuccessful.
The prosecution must establish the relevant professional breach and causal relationship.
Application to therapeutic AI:
If a physician follows an AI-generated diagnosis that results in patient death, investigators should examine whether the physician improperly relied on the algorithm, whether the software was defective and whether the alleged breach caused the outcome.
legalacts.ru
Case 2. Constitutional Court of Russia – 14 January 2020
Decision No. 5-O of 14 January 2020
Subject: Criminal investigation into alleged medical negligence following a patient's death.
The proceedings concerned a patient's death following surgery and disagreements about the appropriate criminal-law classification of the medical professional's alleged conduct.
The case involved the distinction between:
Negligent causing of death through professional misconduct.
Negligence in the performance of official duties.
Provision of services failing to meet safety requirements.
Legal significance:
The legal classification of medical wrongdoing must depend on the actual circumstances and the elements of the relevant offence.
Application to therapeutic AI:
Where an AI-assisted surgical or therapeutic decision causes harm, the court must distinguish a professional clinical error from organisational failures or the release of an unsafe medical product.
ГАРАНТ
Case 3. Second Cassation Court of General Jurisdiction – 4 June 2024
Case No. 8G-10806/2024 (88-13520/2024)
Subject: Medical negligence and compensation for harm following a patient's death.
The proceedings concerned a patient who died after inadequate medical treatment.
The judicial materials discussed failures in clinical assessment, investigations, treatment and the management of the patient's condition.
The civil proceedings resulted in an award of RUB 250,000 in moral-harm compensation to a family member.
Legal significance:
Medical organisations and responsible professionals may face civil consequences where defects in medical care are established.
Application to therapeutic AI:
A hospital cannot automatically avoid responsibility by arguing that a treatment decision was based on software rather than the independent professional judgment of its medical staff.
ГАРАНТ
Case 4. Constitutional Court of Russia – 25 June 2024
Decision No. 1476-O of 25 June 2024
Subject: Negligent medical treatment and the distinction between Articles 109 and 238 of the Criminal Code.
The Constitutional Court considered a complaint concerning the different criminal-law treatment of negligent medical conduct and unsafe services.
The underlying circumstances involved medical workers whose negligent actions allegedly contributed to a patient's death.
The Court explained that the relevant criminal provisions establish different offences and must be applied according to the circumstances of each case.
Legal significance:
The distinction between negligent professional misconduct and unsafe service provision is important when determining criminal liability.
Application to therapeutic AI:
Where AI-assisted treatment causes death, the investigation must identify whether the conduct involved a physician's professional negligence, a wider organisational safety failure or another legally relevant offence.
Документы ленты ПРАЙМ: ГАРАНТ.РУ
Case 5. Supreme Court of the Russian Federation – Plenary Resolution No. 18 of 25 June 2019
Subject: Judicial interpretation of Article 238 of the Criminal Code.
This resolution concerns the judicial application of criminal provisions governing goods, work and services that fail to satisfy safety requirements.
It provides an important interpretative framework for cases involving allegedly unsafe medical services.
Application to therapeutic AI:
If a healthcare organisation systematically deploys unsafe AI software in clinical practice, the question may arise whether the resulting conduct satisfies the requirements of Article 238.
However, a software defect or isolated medical error does not automatically establish this offence.
Верховный Суд Российской Федерации+1
Overall judicial principle: Existing Russian case law supports examining professional standards, causation, patient injury, organisational responsibility and product safety. These precedents are analogies for therapeutic AI disputes, not judgments establishing an AI-specific liability doctrine.
9. Clinical Trials and AI-Based Biotechnology Research
Therapeutic AI can be used in clinical research involving:
Gene therapy.
Stem-cell treatments.
Personalised cancer medicine.
Genetic disease prediction.
AI-assisted pharmaceutical development.
Biological drug-response prediction.
Precision medicine.
Russian clinical investigations must comply with the applicable legislation, including Law No. 61-FZ on the Circulation of Medicines and the medical-device registration framework.
An AI system used to analyse clinical-trial information does not eliminate the need for scientifically valid research, appropriate ethical safeguards and regulatory approval.
For example, an AI model might predict that a particular gene therapy will be effective for a patient with a rare genetic disorder. That prediction cannot independently establish that the therapy is safe or authorised for clinical use.
10. Major Legal Challenges
Several issues remain important for the development of therapeutic AI regulation in Russia.
1. Algorithmic transparency
Patients and regulators may have difficulty understanding how an AI system reached a particular treatment recommendation.
2. Clinical accountability
There must be a clear distinction between the responsibility of the treating physician, hospital and software manufacturer.
3. Algorithmic discrimination
AI systems trained on incomplete or unrepresentative medical datasets may produce less reliable recommendations for particular patient populations.
4. Data security
Genomic and medical information can reveal highly sensitive personal information. Unauthorised disclosure may create serious legal consequences.
5. Software updates
Changes to an AI model may affect its clinical performance. Manufacturers must consider whether software changes require amendments to the medical-device registration documentation.
6. Cross-border biotechnology
Foreign AI developers supplying medical software to Russian healthcare organisations must consider Russian medical-device registration, data protection and applicable import requirements.
7. Regulatory uncertainty
The rapid development of generative AI, autonomous therapeutic systems and personalised biotechnology creates regulatory questions that existing healthcare legislation does not always answer expressly.
11. Hypothetical Legal Case Study
Situation: AI-assisted cancer treatment failure
A Russian hospital uses registered AI medical software to analyse cancer patients' diagnostic images.
The system incorrectly classifies a malignant tumour as benign. The treating physician relies on the recommendation, delays treatment and the patient's condition deteriorates.
The patient subsequently suffers serious injury.
Legal questions:
Was the AI software registered for the relevant medical purpose?
Did the manufacturer adequately validate the software?
Was the software operating within its approved intended purpose?
Did the physician independently assess the patient's condition?
Did the hospital implement appropriate clinical supervision?
Were relevant errors recorded and reported?
Is there a demonstrable causal connection between the software failure, medical decision and injury?
Potential legal consequences:
Civil compensation under the Russian Civil Code.
Professional or disciplinary consequences for responsible medical personnel.
Regulatory investigation by Roszdravnadzor.
Product-related liability against the manufacturer where the legal conditions are satisfied.
Criminal investigation where the elements of a criminal offence are established.
The patient would not need to prove that the AI itself was a legal person. The relevant question would be which legally responsible person or organisation caused or contributed to the injury.
12. Recommendations for Strengthening Therapeutic AI Oversight
A more comprehensive Russian oversight framework could incorporate:
Mandatory documentation of clinically significant AI recommendations.
Independent validation of therapeutic algorithms.
Human review of high-risk treatment decisions.
Clear notification when AI materially influences patient treatment.
Regular auditing of algorithmic accuracy.
Mandatory investigation of serious AI-related medical incidents.
Clear contractual allocation of responsibilities between hospitals and manufacturers.
Stronger protection of genetic and medical information.
Transparent procedures for software updates.
Special regulatory requirements for AI-assisted gene therapy and personalised medicine.
These are proposed regulatory improvements, not a statement that every requirement already exists as an express statutory obligation in Russia.
13. Conclusion
Therapeutic AI oversight in Russia is developing through the interaction of healthcare legislation, medical-device regulation, personal-data protection, pharmaceutical law and civil and criminal liability.
Federal Law No. 323-FZ provides the fundamental patient-protection framework, while Government Resolution No. 1684 of 2024 establishes the current medical-device registration framework.
Particularly significant is Roszdravnadzor Order No. 123 of 10 February 2026, which introduces a specific system for monitoring operational information from qualifying AI-based medical devices.
Existing Russian judicial decisions also provide important principles concerning medical negligence, professional duties, causation and patient compensation.
However, a major legal issue remains unresolved: the precise allocation of responsibility when a defective therapeutic algorithm, a physician's clinical decision and a hospital's organisational failures jointly contribute to patient injury.
The future development of Russian therapeutic AI law will therefore depend on effective regulatory supervision, scientific validation, human clinical accountability and enforceable patient-protection mechanisms.
14. Principal Legal Authorities and Research References
| Legal authority | Relevance |
|---|---|
| Federal Law No. 323-FZ (2011) | Healthcare and patient rights |
| Federal Law No. 152-FZ (2006) | Personal-data protection |
| Federal Law No. 61-FZ (2010) | Medicines and pharmaceutical regulation |
| Federal Law No. 86-FZ (1996) | Genetic engineering |
| Government Resolution No. 1684 (2024) | Medical-device registration |
| Roszdravnadzor Order No. 123 (2026) | AI medical-software monitoring |
| Civil Code Articles 1064, 1068, 1095–1098 | Civil and product liability |
| Criminal Code Articles 109, 118, 238 and 293 | Potential criminal liability |
| Supreme Court Plenary Resolution No. 18 (2019) | Interpretation of Article 238 |
Useful primary and research sources:
Roszdravnadzor – Order No. 123 of 10 February 2026.
Government Resolution No. 1684 – Medical Device Registration Rules.
Roszdravnadzor – Medical Device Registration.
Supreme Court – Plenary Resolution No. 18.
Supreme Court – Case No. 11-UD19-1.
Second Cassation Court – Case No. 8G-10806/2024.
Research note: This is an academic legal overview, with legislation and judicial materials considered as of 18 November 2026. The cited medical cases are relevant analogies rather than direct therapeutic-AI precedents. For litigation or regulatory submissions, consult the current official Russian-language statutory texts and full judgments.

comments