The Central Bank of Russia has issued methodological recommendations to help financial institutions manage information security when using artificial intelligence. The document is the regulator’s first to systematise the risks associated with AI adoption, outline potential cyberattack tactics against AI systems and set out protection measures. The recommendations include a control for high-risk uses of AI in critical business processes. Where AI is used in processes with elevated information security risks, specifically payment transactions, the relevant operation should be confirmed by a human employee. Financial institutions are also advised to develop their own threat models and information security policies for working with AI, with responsibility for preparing those internal documents assigned to the organisation’s deputy head for information security. A separate section covers information security issues linked to vendor-provided AI services, including the view that a vendor AI model’s participation in a bug bounty programme increases trust in that model.