Artificial intelligence is changing the speed and scale of cyber threats facing financial institutions. Attackers may use AI to improve phishing campaigns, automate reconnaissance, generate convincing fraudulent communications, or accelerate parts of an intrusion. For banks, the challenge is not simply to detect these threats. It is to make sound decisions, protect critical services, coordinate teams, and recover operations when an incident unfolds.
AI cyberattack tabletop exercises for banks provide a structured way to test these capabilities without launching a live attack. They allow board members, executives, cybersecurity teams, risk officers, and business continuity leaders to rehearse realistic scenarios and identify weaknesses in existing response plans. This guide explains how to design effective exercises, evaluate their results, and turn lessons into measurable improvements.
Why AI-Enabled Cyberattack Exercises Matter for Banks
On October 16, 2026, Reuters reported that G7 finance ministers and central bank governors had agreed to conduct tabletop exercises to prepare for potential AI-enabled cyberattacks against financial systems. The initiative reflects growing concern about the implications of advanced AI for financial stability and cyber resilience. Read the Reuters report.
For banks, the practical message is clear: cybersecurity preparedness must include the decisions and coordination required during a serious disruption, not just the technical controls intended to prevent one.
A bank may have monitoring systems, incident response procedures, backup infrastructure, and security specialists. However, these capabilities can still fail to work together effectively under pressure. Senior leaders may receive conflicting information, departments may disagree about whether to suspend a service, and external communications may be delayed while teams establish the facts.
A tabletop exercise allows the institution to examine these weaknesses in a controlled environment.
The exercise is particularly valuable because banking incidents can affect more than information systems. They may disrupt customer access, payments, liquidity-related processes, regulatory reporting, third-party services, and public confidence.
What Is an AI Cyberattack Tabletop Exercise?
A tabletop exercise is a facilitated discussion in which participants respond to a simulated incident using their existing policies, responsibilities, and procedures. A facilitator introduces developments in stages, and participants explain what they would do, who would make decisions, and which resources they would use.
An AI-focused exercise applies this approach to a scenario in which artificial intelligence contributes to the threat, the attack's scale, the difficulty of detection, or the speed of escalation.
The objective is not to prove that a bank can predict every possible attack. Instead, it is to determine whether the organization can:
- Recognize and assess a developing incident.
- Establish clear incident ownership and decision authority.
- Coordinate cybersecurity, risk, legal, compliance, operations, and communications teams.
- Protect critical banking services and sensitive information.
- Meet applicable notification and reporting obligations.
- Recover operations safely and learn from the incident.
Unlike a live penetration test, a discussion-based tabletop exercise does not require participants to compromise production systems. It can therefore test strategic and organizational readiness without introducing unnecessary operational risk.
AI-Enabled Cyberattack Scenarios Banks Should Test
Scenario selection should reflect the institution's risk assessment, critical operations, technology environment, and third-party dependencies. A useful exercise combines a credible initial event with developments that force participants to make difficult decisions.
1. AI-Assisted Phishing and Executive Impersonation
A bank receives convincing messages that appear to come from senior management, a trusted supplier, or a financial institution. The communications request urgent payment changes, credentials, or access to sensitive information.
The exercise examines whether staff verify unusual requests, whether identity controls detect suspicious activity, and whether the bank can investigate a potential compromise before fraudulent instructions are acted upon.
Executives should consider who can suspend a payment, freeze an affected account, or authorize additional verification without unnecessarily disrupting legitimate transactions.
2. Automated Attacks Against Customer Accounts
The scenario involves a sudden increase in suspicious login attempts, account takeover indicators, or fraudulent transactions. AI may help attackers adapt their approaches or scale activity across multiple accounts.
Participants must decide how to distinguish malicious activity from legitimate customer behavior, when to introduce additional authentication, and how to protect customers without creating excessive friction.
The exercise should also examine customer support capacity, fraud escalation procedures, and the communication of protective measures.
3. Compromise of a Critical Technology Supplier
A bank's cloud provider, payment processor, software supplier, or managed security service reports a potential security incident. The bank must determine whether its own systems or critical services are affected.
This scenario tests supplier notification procedures, incident ownership, access to independent evidence, contractual escalation requirements, and the availability of alternative operating arrangements.
It is especially important where several financial institutions depend on the same technology provider.
4. Manipulation of AI Systems or Sensitive Data
Participants receive evidence suggesting that an AI-enabled banking application has produced unreliable outputs or that data used by the system may have been exposed or manipulated.
The response should consider whether to restrict the system, preserve evidence, validate affected decisions, notify relevant stakeholders, and maintain essential services through a controlled alternative.
The exercise must distinguish a confirmed compromise from an unverified allegation. Premature assumptions can lead to unnecessary service interruptions or ineffective containment.
5. A Multi-System Incident Affecting Critical Banking Services
The final scenario combines suspicious access, service disruption, uncertain data integrity, and pressure from customers or regulators. Participants must decide whether to isolate affected systems, activate continuity arrangements, prioritize restoration, and communicate what is known.
This scenario tests whether the bank can manage a complex incident as an enterprise-wide operational risk rather than treating it solely as a cybersecurity investigation.
How to Design an Effective Banking Tabletop Exercise
A strong exercise is built around explicit objectives, realistic decisions, and measurable learning outcomes. The following process provides a practical starting point.
Step 1: Define the Critical Services at Risk
Begin by identifying the banking operations that must remain available or be restored within acceptable timeframes.
Depending on the institution, these may include payment processing, customer authentication, account access, treasury operations, settlement activities, and essential reporting systems.
Identify the technology assets, internal departments, external providers, and data flows that support each service. This gives the scenario a realistic operational foundation.
The Basel Committee on Banking Supervision's principles for operational resilience emphasize governance, business continuity, third-party dependency management, incident management, and resilient information and communication technology. Banks can use these principles to inform exercise objectives and recovery priorities.
Explore the Basel Committee's operational resilience guidance.
Step 2: Set Clear Exercise Objectives
Avoid trying to test every aspect of cybersecurity in a single session. Select a manageable set of objectives, such as:
- Testing how quickly an incident is escalated to the appropriate decision-makers.
- Confirming who can authorize service restrictions or emergency recovery measures.
- Assessing coordination between cybersecurity, operations, risk, legal, and compliance functions.
- Testing decisions about customer communication and regulatory notification.
- Evaluating whether critical services can be restored using trusted systems and verified data.
Each objective should correspond to a procedure, responsibility, or capability that can be assessed.
Step 3: Assign Participants and Decision Rights
Invite representatives whose decisions are necessary to manage the scenario. Depending on its scope, the exercise may include the CISO, CIO, COO, CRO, compliance leadership, legal counsel, business continuity managers, fraud teams, communications specialists, and representatives responsible for critical suppliers.
Board members or designated board representatives may participate when the objective includes strategic oversight, risk appetite, major service disruption, or significant reputational consequences.
Before the exercise begins, clarify the difference between making a decision, providing technical advice, approving an action, and communicating its consequences. Ambiguous authority can delay containment and recovery.
Step 4: Develop a Scenario With Progressive Updates
The facilitator should introduce the incident in stages rather than revealing every detail at the beginning.
For example:
- The bank detects unusual authentication activity and receives reports of suspicious customer communications.
- Fraud indicators increase, while the technical team cannot initially confirm the scope of the incident.
- A critical supplier reports a related security concern.
- Customers begin reporting service difficulties, and senior management requests an impact assessment.
- Investigators discover that some records require validation before normal processing can safely resume.
At each stage, participants should explain their decisions, required evidence, escalation route, and expected outcome.
The scenario should be challenging but plausible. It should not assume that every AI-enabled attack succeeds or that all security controls fail simultaneously.
Step 5: Test Decisions, Not Just Knowledge
A tabletop exercise becomes less useful when participants simply describe what a policy says.
Ask them to identify the actual decision they would make, the person authorized to make it, the information needed, and the consequences of acting or waiting.
Useful prompts include:
- What evidence is sufficient to declare a major incident?
- Which services should be restricted, and who approves the restriction?
- How will the bank establish whether critical records remain trustworthy?
- When should customers, regulators, suppliers, and other institutions be notified?
- What conditions must be met before affected systems return to normal operation?
These questions reveal whether written procedures can guide action when facts are incomplete and time is limited.
Step 6: Record Findings and Assign Corrective Actions
An exercise should produce more than a meeting summary. Record the decisions made, unresolved questions, procedural gaps, resource constraints, and dependencies that could delay response or recovery.
Assign each significant finding to a named role or responsible function. Set a target completion date and define what evidence will demonstrate that the issue has been resolved.
Corrective actions might include clarifying escalation authority, updating supplier contact arrangements, validating backup procedures, improving incident communications, or conducting a technical test of a recovery capability that could not be verified during the discussion.
How Banks Should Measure Exercise Effectiveness
A successful tabletop exercise does not necessarily mean that every participant answered every question correctly. Its value lies in revealing whether the bank can make defensible decisions and improve its response capabilities.
Useful evaluation measures include:
- Escalation clarity: Did participants identify the correct decision-makers and escalation routes?
- Decision quality: Were decisions consistent with documented policies, risk appetite, and the available evidence?
- Cross-functional coordination: Did technical, operational, legal, compliance, and communications teams share information effectively?
- Critical-service protection: Were service priorities and acceptable disruption limits clearly understood?
- Recovery readiness: Could participants identify the resources, dependencies, evidence, and approvals needed to restore operations safely?
- Corrective-action completion: Were weaknesses assigned, tracked, and verified after the exercise?
Banks can record elapsed time between scenario updates and key decisions as a diagnostic measure. However, the exercise should not treat speed alone as success. A rapid decision based on incomplete evidence may increase operational or customer harm.
Metrics should be interpreted in the context of scenario complexity, institutional size, applicable requirements, and the consequences of each decision.
Governance, Regulation, and Operational Resilience
AI-related cyber exercises should be integrated into the bank's wider risk management and business continuity arrangements.
The Basel Committee's operational resilience principles emphasize response and recovery planning, testing, third-party dependencies, incident management, and learning from previous events. The Bank for International Settlements' operational resilience summary provides a useful reference for these areas.
The NIST Cybersecurity Framework can also help organizations structure cybersecurity risk management across governance, identification, protection, detection, response, and recovery.
These frameworks support the design of a coherent exercise programme, but they do not replace jurisdiction-specific banking rules. Institutions must determine which reporting deadlines, supervisory expectations, data protection obligations, and incident notification requirements apply to their activities and locations.
Senior management should also ensure that exercise findings reach the appropriate governance bodies. Material weaknesses in critical-service recovery, decision authority, or third-party resilience should not remain confined to the cybersecurity team.
Common Mistakes to Avoid
Several weaknesses can limit the value of tabletop exercises.
Focusing Exclusively on Technical Teams
Cybersecurity specialists may understand the threat but cannot independently resolve every decision involving customer services, legal obligations, risk acceptance, or public communication. Include the functions needed to manage the full incident lifecycle.
Using Unrealistic Scenarios
A scenario that assumes every control fails provides limited insight into realistic readiness. Build a plausible sequence of events and allow existing controls to work where appropriate.
Treating Participation as Proof of Preparedness
Attendance does not demonstrate that procedures are effective. Require participants to explain decisions, responsibilities, evidence requirements, and recovery conditions.
Ignoring Suppliers and External Dependencies
Banks rely on external technology, data, communications, and payment services. Exercises should test how the institution would respond if a supplier could not provide timely information or restore a critical service.
Failing to Verify Improvements
Repeated exercises have limited value when previous findings remain unresolved. Track corrective actions and retest important weaknesses to determine whether the bank's capabilities have improved.
Executive Checklist for AI Cyberattack Readiness
Before approving a tabletop exercise, banking leaders should confirm that:
- The scenario reflects the institution's critical services and credible AI-related cyber risks.
- Exercise objectives are specific and linked to existing response or recovery plans.
- Decision authority and escalation responsibilities are clearly defined.
- Cybersecurity, operations, risk, legal, compliance, communications, and relevant supplier functions are represented.
- Customer protection, data integrity, service continuity, and recovery decisions are included.
- Regulatory and contractual notification requirements are considered.
- Findings will be assigned, tracked, and independently verified where appropriate.
- Material weaknesses will be reported through the appropriate governance channels.
- Follow-up exercises or technical tests will verify progress.
This checklist should be adapted to the bank's risk profile, operating model, and applicable supervisory requirements.
Questions Banking Leaders Should Ask
How often should banks conduct cyberattack tabletop exercises?
The frequency should reflect the institution's risk profile, significant technology or supplier changes, previous incidents, and applicable regulatory expectations. Banks should also conduct additional exercises when major weaknesses emerge or response plans change substantially.
Should board members participate directly?
Board participation is useful when the exercise tests strategic oversight, risk appetite, major service disruption, crisis governance, or decisions reserved for the board. Other exercises may use executive participants and provide the board with a concise summary of significant findings.
How are AI-focused exercises different from traditional cyber exercises?
AI-focused exercises examine how AI may influence the scale, speed, credibility, or uncertainty of an incident. They can also consider compromised AI applications, unreliable outputs, data manipulation, and questions about human oversight. Core incident response and recovery principles remain essential.
Can a tabletop exercise prove that a bank is cyber resilient?
No. A discussion-based exercise identifies decision-making and procedural weaknesses, but it does not prove that technical controls work. Banks should combine tabletop exercises with appropriate technical testing, recovery validation, access-control reviews, and supplier assessments.
What should happen after an exercise?
The bank should document findings, assign accountable owners, set deadlines, prioritize material risks, and verify that corrective actions work. Significant lessons should inform updates to response plans, training, governance, and subsequent exercises.
Conclusion
AI cyberattack tabletop exercises for banks provide a practical way to test whether cybersecurity arrangements translate into effective executive decisions and operational recovery. Their value comes from realistic scenarios, clear responsibilities, evidence-based decisions, and disciplined follow-up.
Banks should treat each exercise as part of a continuing resilience programme. By connecting cybersecurity, governance, business continuity, supplier management, and executive oversight, institutions can identify weaknesses before a real incident forces them to respond under pressure.
The next step is to select one critical banking service, define a credible AI-related threat scenario, and test the decisions required to protect and restore that service.
Related Professional Development
Banking professionals who want to strengthen their understanding of AI applications, governance, and risk considerations can explore Gentex Training Center's Artificial Intelligence (AI) in Banking course. The programme provides a relevant professional development pathway for understanding AI's role in banking and the considerations involved in its responsible application.
Explore the course to identify learning opportunities that align with your institution's technology, risk management, and governance priorities.
About the Author
Yusef
Banking Cyber Resilience Expert at Gentex Training