Dataset Viewer
Auto-converted to Parquet Duplicate
Taxonomy
stringclasses
15 values
Scenario
stringlengths
7
31
Risk Description
stringlengths
30
122
Priority
int64
1
3
Control Name
stringlengths
9
40
Control Activities
stringlengths
58
162
COBIT Objectives
stringclasses
5 values
Attribution
stringclasses
1 value
Related Standards
stringclasses
1 value
Strategy
Strategy deficiency
Wasted resources and reputational damage due to lack of clear enterprise-wide AI strategy.
1
Enterprise AI strategy
Develop and implement a comprehensive enterprise-wide AI strategy aligning with business objectives. Establish governance structures, roles, and responsibilities.
Align, Plan and Organize
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Strategy
Misaligned strategy
Missed opportunities due to insufficient stakeholder engagement, leading to AI systems that do not support business goals.
1
Stakeholder alignment
Establish stakeholder engagement processes. Develop communication protocols to ensure continuous alignment between AI projects and business objectives.
Monitor, Evaluate and Assess
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Governance
Misaligned ethics
Reputational damage due to AI decisions conflicting with organizational ethical values.
1
Responsible AI principles
Develop AI ethics guidelines, establish an ethics review board, and integrate ethical considerations into design and deployment.
Monitor, Evaluate and Assess
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Governance
Overconfidence in automation
Wasted resources due to unrealistic expectations and overconfidence in AI capabilities.
1
Capability and limitations communication
Communicate limitations and potential risks of AI. Avoid overstating capabilities to ensure realistic expectations.
Evaluate, Direct and Monitor
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Governance
Governance erosion
Weakened control due to AI impacts on existing governance, leading to reduced control over data processing.
1
Control integration
Update governance frameworks to incorporate AI-specific considerations. Ensure alignment with established policies and risk protocols.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Governance
Compliance failure
Regulatory penalties and reputational damage due to non-compliance with AI regulations.
1
Compliance audit
Regularly test, audit, monitor and assess control compliance of AI systems with internal policies and external regulations.
Monitor, Evaluate and Assess
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Governance
Vendor lock-In
Limited flexibility due to lack of exit strategies, leading to dependence on single vendors.
1
Exit planning
Include exit strategy considerations in design and procurement. Ensure ability to migrate to alternative providers.
Align, Plan and Organize
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Governance
Trust deficit
Limited innovation due to lack of trust in AI technologies, restricting deployment.
2
AI framework
Document capabilities, provide clear explanations of decisions, and establish processes for independent verification.
Align, Plan and Organize
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Governance
Communication breakdown
Ineffective governance due to lack of common AI language, leading to misunderstandings.
2
AI glossary management
Develop and maintain a glossary of AI terms. Ensure consistent use of terminology across the organization.
Evaluate, Direct and Monitor
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Governance
Ownership vacuum
Unauthorized development and security breaches due to lack of defined ownership.
2
Operating model definition
Establish clear roles, responsibilities, and accountabilities for AI initiatives and ensure segregation of duties.
Evaluate, Direct and Monitor
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Decision-making
Risk integration gap
Unmanaged AI risks due to failure to integrate risk assessment into the AI framework.
1
Mandatory risk and impact assessments
Mandate systematic identification, evaluation, and management of AI risks. Include quantitative model impact assessments.
Evaluate, Direct and Monitor
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Decision-making
AI model risk exposure
Inadequate quantitative model risk management leading to inaccurate models and financial losses.
1
Quantitative risk model
Implement quantitative model risk management. Conduct impact assessments and statistical analyses of model accuracy.
Align, Plan and Organize
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
People
Resource misalignment
Staffing inefficiencies due to unclear resourcing requirements in AI strategy.
1
Staff planning
Define and document human resource requirements including recruitment, role profiles, and training.
Evaluate, Direct and Monitor
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
People
Talent flight
Ineffective AI systems due to poor development and retention of human talent.
1
Talent alignment
Establish HR processes to recruit, develop, and retain talent aligned with AI strategy.
Align, Plan and Organize
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
People
Knowledge deficit
Ineffective operations and poor incident response due to insufficient knowledge retention.
1
Knowledge continuity
Assign individuals to sustainment functions. Ensure knowledge retention through formal management practices.
Align, Plan and Organize
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Architecture
Unexplainability
Inability to explain AI decisions due to lack of interpretability functionality.
1
Explainability by design
Integrate explainability into requirements. Ensure explanations are clear and accessible. Document explainability features.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Architecture
Incompatibilities
Operational issues due to integration, scalability, and compatibility failures.
1
Compatibility testing
Perform compatibility testing for software, hardware, and networks. Conduct scalability assessments and stress testing.
Align, Plan and Organize
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Architecture
Misaligned architecture
Inefficient automation due to undefined AI architecture preventing unified solutions.
2
Architecture alignment
Define enterprise AI architecture including preferred technologies, security controls, and monitoring requirements.
Align, Plan and Organize
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Architecture
Segregation deficiency
Security/integrity losses due to lack of architectural segregation in cloud/multi-tenant envs.
2
Architectural segregation
Enforce segregation of AI system components and data from other infrastructure. Ensure data integrity via audits.
Align, Plan and Organize
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Architecture
Unavailability
Business disruption due to insufficient availability of AI systems.
2
High availability
Implement architecture for resilience. Define availability metrics, redundancy plans, and reliability testing.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Architecture
Ineffective security
Vulnerabilities due to failure to embed 'security by design' principles.
3
Security by design
Embed security policies in architecture and development. Ensure adherence to security standards.
Align, Plan and Organize
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Architecture
License noncompliance
Legal/financial losses due to non-compliance with software licenses.
3
License management
Establish license management system. Track usage, renewals, and compliance with vendor requirements.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Lifecycle
Poor hypothesis
Unreliable AI outcomes due to poor hypothesis quality and inadequate governance.
1
Hypothesis testing
Require approval of hypotheses based on predefined criteria. Regular reviews for relevance.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Lifecycle
Poor algorithms
Ineffective performance due to poor algorithm quality and inadequate controls.
1
Algorithmic controls
Enforce governance policies for algorithm development. Document standards and validation procedures.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Lifecycle
Flawed logics
Unreliable outputs due to flawed logic or inaccurate model parameters.
1
Logic validation
Establish rigorous logic validation procedures. Document logic and parameters. Require approval for changes.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Lifecycle
Data accuracy failures
Inaccurate data resulting in erroneous outputs due to failure to maintain standards.
1
Data accuracy verification
Enforce data accuracy verification standards. Document validation, error detection, and correction processes.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Lifecycle
Unallocated roles
Regulatory/operational failures due to unclear data governance roles.
2
Data ownership
Establish roles for data governance (owners, stewards). audit adherence to roles.
Align, Plan and Organize
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Lifecycle
Data corruption
Business disruption due to data corruption from unintended system interactions.
2
Data integrity monitoring
Monitor data interactions. Log all interactions and alert on corruption. Require approval for source changes.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Lifecycle
Integration failure
Operational failure due to corrupted inputs/outputs from external system interactions.
2
Integration testing
Enforce integration testing for inputs/outputs. Monitor for anomalies and remediate corrupted data.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Lifecycle
Poor data
Ineffective AI systems due to poor quality of learning or production data.
2
Data governance
Implement quality checks on learning and production data. Document cleaning procedures and approvals.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Lifecycle
Incomplete inputs
Incorrect outcomes due to use of inaccurate or incomplete data.
2
Data completeness control
Ensure completeness of inputs. Document selection and validation processes. Audit data integrity.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Lifecycle
Inaccurate results
Losses due to models not accurately reflecting underlying parameters.
3
Model validation
Implement validation protocols including sensitivity analysis and benchmarks. Monitor for model drift.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Project
Operational misalignment
Misalignment with organizational goals due to lack of strategic alignment.
2
Business case
Establish framework to ensure AI initiatives support goals. Formal approval of AI business cases.
Align, Plan and Organize
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Project
Problem mismatch
Financial losses due to mismatch between business problem and model design.
2
Iterative development
Adopt iterative approaches (Agile). Use prototyping to identify mismatches early.
Align, Plan and Organize
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Project
Management gap
Failure to meet requirements due to poor project management.
2
Program management
Implement program management methodology. Track resources, timelines, and stakeholder engagement.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Project
Poor benefits
Unmet strategic goals due to failure to track ROI and benefits.
3
Impact value
Develop benefits management framework. Track metrics for short/long term benefits realization.
Evaluate, Direct and Monitor
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Project
Assurance deficit
Project failures due to lack of independent assurance over objectives.
3
Assurance
Engage independent assurance function. Report on quality, costs, compliance, and controls.
Evaluate, Direct and Monitor
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Development
Design flaw
Unsuccessful delivery due to poor design/development methodology.
1
Development standards
Establish AI development standards integrated with broader IT standards. Ensure consistent application.
Align, Plan and Organize
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Development
Inaccurate model
Losses due to lack of detailed model universe definition.
1
Model universe
Define the AI model universe (data sources, quality, assumptions). Update regularly.
Align, Plan and Organize
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Development
Variable misalignment
Incorrect results due to inappropriate variable relationships (correlation vs causality).
1
Relationship modeling
Ensure relationships between variables and hypotheses are defined correctly. Distinguish correlation/causality.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Development
Overfitting
Model performs well on training data but poorly on new data due to complexity.
1
Overfitting mitigation
Design algorithms to prevent overfitting. Document designs and test on unseen data.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Development
Learning bias
Discriminatory outcomes due to data bias in learning data.
1
Bias mitigation
Implement controls for sensitivities (race/gender/etc). Document evaluation and bias checks.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Development
To-be inaccuracy
Failure to meet needs due to poor knowledge of To-Be processes.
2
To-be analysis
Maintain documentation of user stories and process flows. Ensure approval by sponsors.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Development
Insufficient segregation
Failures during deployment due to insufficient testing in production-like environments.
2
Environment segregation
Maintain separate Dev, QA, and Prod environments. Test in environment identical to production.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Development
Temporal misalignment
Incorrect insights due to misalignment of data time scales.
2
Synchronization verification
Ensure sourced data aligns with AI system time scale. Document synchronization processes.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Development
Data duplication
Incorrect insights/wasted resources due to processing duplicate data.
2
Duplication mitigation
Prevent processing of same data multiple times. Implement file/data validation checks.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Development
AutoML issues
Model complexity/lack of transparency due to Automated ML usage.
2
AutoML guides
Develop guidelines for AutoML usage. Assess complexity vs performance. Integrate explainability tools.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Development
As-is ignorance
Suboptimal automation design due to poor knowledge of As-Is processes.
3
As-is analysis
Document pre-automation process narratives/flowcharts including exceptions.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Development
Undefined controls
Increased vulnerabilities due to undefined risks and controls.
3
Control matrix
Maintain risk-based control matrix covering solution, interfaces, and management processes.
Align, Plan and Organize
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Development
Control gap
Operational risks due to controls not being implemented in the solution.
3
Control testing
Test that internal controls are developed/configured in line with design.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Development
Weak traceability
Audit challenges due to bot activities not being traceable.
3
Bot identification
Assign unique identifiers (e.g., static IP) to bots to trace activities.
Align, Plan and Organize
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Development
Improper testing
Operational disruptions due to insufficient testing and go-live strategy.
3
Testing execution
Define testing and go-live strategy including migration and contingency. Collect evidence of testing logs.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Development
User acceptance deficiency
Misaligned solutions due to inadequate UAT without business input.
3
Test approvals
Conduct UAT with business input. Maintain records of feedback and approvals.
Align, Plan and Organize
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Operations
Resource laxity
Inefficient resource use/costs due to inadequate control over unpredictable AI demands.
2
Project monitoring
Monitor IT resource demands closely. Prevent inefficient use and increased costs.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Operations
Error oversight
Unauthorized changes/data loss due to undetected system errors.
2
Incident management
Establish consistent incident management procedures. Ensure timely resolution and escalation.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Operations
Undetected error
Delayed incident resolution due to lacking procedures.
2
Error resolution
Perform timely processing of exceptions. Monitor for interruptions and completeness.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Operations
Unsupported jobs
Delayed resolution due to inconsistent job monitoring.
2
Job monitoring
Implement job monitoring to ensure completeness/timeliness. Identify interruptions.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Operations
Capacity issues
Inability to meet demand due to inadequate capacity management.
2
Capacity management
Implement availability/capacity management. Embed scalability in design.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Operations
Performance drift
Accuracy degradation due to lack of stability checks and drift monitoring.
2
Stability monitoring
Implement stability checks. Detect data/concept drift. Ensure ongoing validation.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Operations
Shadow AI
Misalignment with strategy due to unknown AI solutions (Shadow IT).
3
AI inventory
Maintain up-to-date inventory of all AI platforms/solutions and dependencies.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Operations
IP loss
Theft of IP/Code held by third parties.
3
IP protection
Establish repository of IP. Secure backups. Mitigate risk in external interactions.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Operations
AI component blindness
Ineffective controls due to lacking understanding of IT components.
3
Configuration management
Maintain CMDB to understand components and relationships. Update via change management.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Monitoring
Outcome blindness
Not achieving results due to insufficient monitoring of behavior against ethics/business needs.
1
Outcome monitoring
Regularly review outcomes using analytics. Ensure alignment with ethical requirements. Maintain audit trails.
Monitor, Evaluate and Assess
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Monitoring
Monitoring ineffectiveness
Reduced effectiveness due to inadequate monitoring of AI operations.
1
Operational monitoring
Real-time monitoring/alerting for anomalies. Detailed logging. Establish KPIs/KRIs.
Monitor, Evaluate and Assess
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Monitoring
Undetected issues
Financial/compliance losses due to lack of post-deployment monitoring.
1
Post-deployment monitoring
Framework for metrics/thresholds. Automated monitoring tools for anomalies. Review KPIs regularly.
Monitor, Evaluate and Assess
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Monitoring
Control override
Financial loss if automated stop/loss controls fail.
2
Automated stop/loss
Design stop/loss controls. Implement override process for exceptions with checks.
Monitor, Evaluate and Assess
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Security
Lack of auditability
Inability to validate outcomes due to unsecured storage.
1
Auditability
Securely store data/algorithms. Comply with privacy/retention regulations. 'Vault principle'.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Security
Unauthorized access
Losses due to inappropriate changes to learning data.
2
Data access
Securely configure input datasets. Prevent unauthorized changes. Validate data input.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Security
Security breach
Data loss due to inconsistent security management.
3
Cyber security
Consistent security management across code/infrastructure/data. Align with ISO 27001.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Security
Malware attack
Harm due to undetected malware in AI environment.
3
Malware protection
Malware protection systems. Monitor patches. Protect self-learning components.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Security
Data breach
Data loss due to insecure temporary files.
3
Encryption
Encrypt code, storage, and network communications. Document protocols.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Security
Vulnerability blindness
Losses due to undetected weaknesses.
3
Vulnerability testing
Periodic penetration tests and Red-team reviews. Test sensors/monitoring.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Access
Compromised bot
Unauthorized actions due to inadequate tracking of Bot accounts.
3
Bot control
Assign unique accounts to Bots. Track usage. Compensating controls for shared accounts.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Access
Compromised user
Tampering with logic due to inadequate access controls.
3
User control
Unique accounts for humans. Restrict access to logic/algorithms.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Access
Excessive bot access
Unauthorized access due to excessive rights assigned to Bots.
3
Bot access authorization
Assign Bot rights on 'need to have' basis. Constrain to specific transactions.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Access
Excessive user access
Unauthorized access due to excessive rights assigned to users.
3
User access authorization
Assign user rights on 'need to have' basis. Constrain access to Bot environment.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Access
Inadequate account provisioning
Unauthorized access due to lack of formal approval.
3
Account provisioning
Procedures for creating accounts/assigning privileges. Require formal business approval.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Access
Inadequate access revocation
Unauthorized access due to failure to delete accounts.
3
Access revocation
Timely deletion/locking of accounts when no longer needed.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Access
Undetected bot access
Unauthorized access due to lack of review of Bot privileges.
3
Bot access review
Periodic reviews of Bot access/privileges. Act on exceptions.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Access
Undetected user access
Unauthorized access due to lack of review of user privileges.
3
User access review
Periodic reviews of user access/privileges. Act on exceptions.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Access
Excessive privileged access
Unauthorized access due to excessive privileged rights.
3
Privileged access authorization
Restrict powerful accounts to admins. Use secure networks/encryption/MFA.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Access
Segregation of duties conflicts
Unauthorized access due to conflicting access rights.
3
Access segregation
Ensure SoD compliance. Prevent single individuals from creating accounts AND approving changes.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Access
Weak authentication
Unauthorized access due to weak passwords/biometrics.
3
Authentication
Strong auth controls (MFA, password complexity) for all accounts including automation.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Access
Unauthorized third-party access
Unauthorized access due to inadequate third-party controls.
3
Third-party control
Subject third parties to same controls as regular users. Monitor access.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Change management
Collateral IT impact
Disruptions to other services due to AI changes.
1
IT Impact assessment
Mandate impact analysis for all changes. Require integration testing.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Change management
Inadequate ongoing testing
Missed defects due to insufficient post-implementation testing.
1
Testing protocol
Establish comprehensive testing protocols. Pre- and post-implementation tests.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Change management
Undetected errors
Operational failures due to lack of automated monitoring for changes.
2
Automated error monitoring
Deploy automated tools to test/validate after changes. Detect errors in real-time.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Change management
Unauthorized changes
Disruptions/vulnerabilities due to lack of change management.
3
Change management
Enforce formal change management. Document risk assessment and approvals.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Change management
Untracked modifications
Performance degradation due to untracked modifications.
3
Modification logging
Log all changes in real-time. Reconcile logs with change records.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Change management
Poor change
Degraded performance due to poor change control.
3
Change control
Strict change control procedures. Review, test, and approve all changes.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Change management
Insufficient validations
Unreliable outcomes due to insufficient validation of changes.
3
Validation procedure
Require pre-deployment tests (functional/security/performance). Document results.
Build, Acquire and Implement
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Third-party
Black box solution
Disruption due to insufficient understanding of logic/IP.
2
Contract review
Define IP ownership and escrow agreements. Ensure right to audit.
Align, Plan and Organize
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Third-party
Third-party default
Losses due to vendors with low control maturity.
3
Due diligence
Subject 3rd parties to same controls or external assurance (ISAE 3402).
Align, Plan and Organize
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Third-party
Third-party dependency
Disruption due to single supplier concentration risk.
3
Third-party segmentation
Categorize suppliers by criticality. Implement contingency plans and backup suppliers.
Align, Plan and Organize
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Third-party
Shadow third-party
Control failures due to unknown vendors.
3
Third party management
Maintain inventory of all 3rd party providers. Integrate into risk/compliance planning.
Align, Plan and Organize
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Business continuity
Inability to recover
Losses due to inability to roll back to last known good state.
1
Roll-Back
Establish roll-back mechanism. Ensure availability of cleansed data for retraining.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Business continuity
Ineffective backhups
Losses due to lack of backups.
1
Backup restoration
Maintain backups/snapshots of AI system and learning data. Test restoration.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Business continuity
Ineffective fallback
Disruptions due to lack of alternative processing facilities.
2
Fallback facility
Establish alternative IT infrastructure and operational capabilities.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Business continuity
Fragility
Prolonged disruption due to lack of continuity plans.
3
Continuity planning
Develop BCP aligned with ISO 22301. Integrate with enterprise BCM.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
Business continuity
Ineffective response
Ineffective response due to lack of testing/training.
3
Continuity testing
Perform regular BCP simulations. Train staff for crisis conditions.
Deliver, Service and Support
Prof. Hernan Huwyler
ISO/IEC 42001, ISO/IEC 42005, COBIT 2019
YAML Metadata Warning: The task_categories "risk-assessment" is not in the official list: text-classification, token-classification, table-question-answering, question-answering, zero-shot-classification, translation, summarization, feature-extraction, text-generation, fill-mask, sentence-similarity, text-to-speech, text-to-audio, automatic-speech-recognition, audio-to-audio, audio-classification, audio-text-to-text, voice-activity-detection, depth-estimation, image-classification, object-detection, image-segmentation, text-to-image, image-to-text, image-to-image, image-to-video, unconditional-image-generation, video-classification, reinforcement-learning, robotics, tabular-classification, tabular-regression, tabular-to-text, table-to-text, multiple-choice, text-ranking, text-retrieval, time-series-forecasting, text-to-video, image-text-to-text, visual-question-answering, document-question-answering, zero-shot-image-classification, graph-ml, mask-generation, zero-shot-object-detection, text-to-3d, image-to-3d, image-feature-extraction, video-text-to-text, keypoint-detection, visual-document-retrieval, any-to-any, video-to-video, other

AI Risk Scenarios & Control Library

Dataset Summary

This dataset contains 100 Common AI Risk Scenarios paired with specific Control Activities. It is designed to serve as the operational backbone for an ISO/IEC 42001 (AI Management System).

While ISO 42001 provides the "What" (Requirements), this dataset provides the "How" (Scenarios and Controls), specifically mapping to ISO/IEC 42005 (AI Risk Management) impact assessments and COBIT 2019 objectives.

Author & Attribution

This library was developed and curated by: Prof. Hernan Huwyler, MBC, CPA

  • Academic Director
  • AI GRC Director

This dataset reflects a synthesis of global best practices in IT Audit and AI Governance.

Dataset Structure

The dataset contains the following fields:

  • Taxonomy: The domain of the risk (e.g., Strategy, Governance, Architecture, Lifecycle).
  • Scenario: The short name of the risk event.
  • Risk description: A detailed explanation of what goes wrong and the resulting impact (Financial, Reputational, Operational).
  • Priority: Suggested risk rating (1 = High, 3 = Low) to assist in heatmapping.
  • Control name: The title of the mitigation strategy.
  • Recommended control activities: The specific steps, documentation, or technical implementations required to mitigate the risk.
  • COBIT 2019 Objectives: Mapping to the COBIT framework (e.g., Align, Plan and Organize).

Use Cases

1. ISO 42001 Implementation (Annex A)

When implementing the internal controls listed in ISO 42001 Annex A, use this dataset to flesh out the specific activities.

  • Annex A Reference: A.6.1 AI System Life Cycle.
  • Dataset Mapping: Filter by Taxonomy = Lifecycle to find controls for "Hypothesis Testing" and "Model Validation."

2. AI Risk Assessment (ISO 42001,ISO 42005, ISO 32894, NIST AI RISK)

Use the Risk description column to populate your Risk Register.

  • Scenario: "Model Overfitting".
  • Impact: "Model performs well on training data but poorly on new data."
  • Mitigation: Implement "Overfitting mitigation" controls defined in this dataset.

3. Audit & Assurance

Internal Auditors can use the Recommended control activities as a checklist to verify if the organization's AI governance is effective.

Example Data

Taxonomy Scenario Priority Control Name Control Activities
Strategy Strategy deficiency 1 Enterprise AI strategy Develop and implement a comprehensive enterprise-wide AI strategy aligning with business objectives.
Lifecycle Data accuracy failures 1 Data accuracy verification Enforce data accuracy verification standards. Document validation, error detection, and correction processes.
Security Vulnerability blindness 3 Vulnerability testing Conduct periodic penetration tests and 'Red-team' reviews.

Citation

If you use this dataset in research or corporate frameworks, please cite:

Huwyler, H. (2024). AI Risk Scenarios & Control Library. Hugging Face Datasets.

Downloads last month
6

Space using hewyler/AIRiskScenarios 1

Collection including hewyler/AIRiskScenarios