On September 9, 2026, the International Technology Congress hosted a roundtable discussion on "Human-Centered AI: How to create an ethical and safe tool for the humanitarian sector?" organized by the International Committee of the Red Cross. The roundtable discussion focused on ethics and safety of using artificial intelligence in humanitarian settings.
The roundtable discussion was attended by representatives of the International Committee of the Red Cross, as well as companies in the sphere of cybersecurity, IT, and representatives of leading Russian universities.
Kirill Chernovol, Researcher of the International Best Practices Analysis Department, spoke at a roundtable discussion on new trends in AI regulation in Russia, with a focus on the technology's application in the humanities. Since September 2026, the Federal Law on the Development of AI Systems has been in effect in Russia, creating a legal basis for selecting and applying large-scale, general-purpose AI models.
The Russian law enshrines principles of human rights protection, respect for human autonomy, and a risk-based approach to the requirements for developers and the operation of AI systems. In this respect, it echoes the International Committee of the Red Cross's policies on the use of AI.
For humanitarian organizations, this means the ability to assess the risks of applying the model to a specific task and implement the necessary protective measures. However, the risk categories are not currently defined in the Russian law. This contrasts with international practice, for example, in the EU, AI-based systems for assessing eligibility for state aid and determining priority of emergency response are classified as high-risk and are subject to increased transparency and risk management requirements.
In Russia, regulators have yet to define these for humanitarian purposes. Kirill Chernovol emphasized that Russia needs to develop regulations for using AI in the humanitarian sector: establish mandatory regular testing of AI systems for errors and risks of discrimination prior to their implementation, as well as procedures for monitoring and auditing AI performance, ensuring transparency and explainability of decisions, and the ability to promptly review AI decisions.