Compliance/Frameworks
ISO/IEC 42001:2023 Controls: Annex A (AI Management System)
AI Governance Controls organised by domain | Reference: ISO/IEC 42001:2023 Annex A
Legend: Critical High Medium Low
A.2: AI-Related Policies
| Code | Title | Priority | Description |
|---|---|---|---|
| A.2.2 | AI policy | High | Define and communicate an AI policy that reflects the organisation's objectives and principles |
| A.2.3 | AI-related roles and responsibilities | High | Assign specific AI governance responsibilities to roles and individuals |
A.3: Internal Accountability for AI
| Code | Title | Priority | Description |
|---|---|---|---|
| A.3.2 | AI system risk assessment process | Critical | Establish a systematic process to identify, analyse and evaluate AI risks |
| A.3.3 | Risk treatment across the AI system life cycle | High | Implement and document risk treatment measures throughout the system life cycle |
A.4: AI Impact Assessment
| Code | Title | Priority | Description |
|---|---|---|---|
| A.4.2 | AI objectives and planning to achieve them | High | Establish measurable AI objectives aligned with the organisational policy and strategy |
| A.4.3 | AI system impact assessment process | Critical | Assess the impacts of AI systems on people, society and the environment before deployment |
A.5: AI Resource Management
| Code | Title | Priority | Description |
|---|---|---|---|
| A.5.2 | Human oversight of AI systems | Critical | Ensure adequate human oversight during the operation of AI systems |
| A.5.3 | Resources for AI systems | Medium | Identify and provide the resources needed to develop and operate AI systems safely |
A.6: Impact Assessment for AI
| Code | Title | Priority | Description |
|---|---|---|---|
| A.6.1 | AI system impact assessment | Critical | Carry out impact assessments for AI systems with potential for significant impact |
| A.6.2 | Societal considerations for AI systems | High | Assess and mitigate the broader societal impacts of AI systems |
A.7: AI System Operations
| Code | Title | Priority | Description |
|---|---|---|---|
| A.7.2 | AI system requirements specification | High | Define functional and non-functional requirements, including security, reliability and ethics |
| A.7.3 | AI system design | High | Ensure that the design incorporates responsible AI principles from the outset |
| A.7.4 | Data for AI systems | Critical | Manage the data life cycle (acquisition, processing, quality, provenance) for AI systems |
| A.7.5 | AI system operations | High | Monitor and manage AI systems in operation to ensure expected behaviour |
| A.7.6 | AI system maintenance | High | Maintain AI systems to preserve performance, security and alignment with objectives |
| A.7.7 | AI system decommissioning | Medium | Plan and carry out the secure decommissioning of AI systems and their data |
A.8: Data for AI Systems
| Code | Title | Priority | Description |
|---|---|---|---|
| A.8.2 | Data acquisition and processing | High | Define and apply processes for the acquisition, preparation and processing of AI data |
| A.8.3 | Data quality | Critical | Establish and monitor quality requirements for data used in training and operation |
| A.8.4 | Data provenance | High | Document and verify the origin and chain of custody of data used in AI systems |
A.9: Information for Interested Parties
| Code | Title | Priority | Description |
|---|---|---|---|
| A.9.2 | Communication with interested parties about AI systems | High | Communicate clearly and transparently with interested parties about AI systems |
| A.9.3 | Documentation for users | High | Provide clear documentation on the functionality, limitations and appropriate use of AI systems |
| A.9.4 | Data management documentation | Medium | Document data management processes for transparency and accountability |
| A.9.5 | Transparency about AI system capabilities and limitations | High | Disclose the capabilities and limitations of AI systems honestly and accessibly |
A.10: Use of AI by the Organisation
| Code | Title | Priority | Description |
|---|---|---|---|
| A.10.2 | Intended use and misuse of AI systems | High | Define intended use and identify misuse risks, including adversarial use |
| A.10.3 | Responsible use of AI systems | Critical | Implement safeguards to ensure that AI systems are used ethically and responsibly |
Total: 25 Annex A controls
Source: ISO/IEC 42001:2023 Annex A, AI Management System