Concepts derived from 'Software Bill of Materials (SBOM) for Artificial Intelligence - Minimum Elements', written by the G7 Cybersecurity Working Group and published by BSI (Federal Office for Information Security, Germany) on 2026-05-12. Categories: Metadata (MD), System Level Properties (SLP), Models (MDL), Dataset Properties (DP), Infrastructure (INF), Security Properties (SP), Key Perf Indicators (KPI).
| ID | Label | Notation | Type | Definition |
|---|---|---|---|---|
dataset-prop | Dataset Properties | G7AI-DP | Elements describing datasets used to train, validate or test AI models. | |
dataset-name | Dataset name | BOM-G7AI-DP-001 G7AI-DP-01 | Requirement | The human-readable name of a dataset used to train, validate or test the AI model. |
dataset-description | Dataset description | BOM-G7AI-DP-002 G7AI-DP-02 | Requirement | A free-text description of the dataset's contents, scope and purpose. |
dataset-content | Dataset content | BOM-G7AI-DP-003 G7AI-DP-03 | The type(s) of data in the dataset (e.g. image, text, audio, tabular) and its size. | |
dataset-identifier | Dataset identifier | BOM-G7AI-DP-004 G7AI-DP-04 | Requirement | A stable, unique identifier for the dataset (e.g. DOI, SPDX ID, dataset registry URL). |
dataset-hash | Dataset hash | BOM-G7AI-DP-005 G7AI-DP-05 | A cryptographic hash of the dataset archive or snapshot for integrity verification. | |
dataset-provenance | Dataset provenance | BOM-G7AI-DP-006 G7AI-DP-06 | Information about how, when and where the dataset was collected, updated and pre-processed. | |
dataset-statistical-prop | Dataset statistical property | BOM-G7AI-DP-007 G7AI-DP-07 | Statistical characteristics of the dataset including size, noise levels, known biases and class distributions. | |
dataset-sensitivity | Dataset sensitivity | BOM-G7AI-DP-008 G7AI-DP-08 | Requirement | The sensitivity classification of the dataset, including whether it contains personal or sensitive data and its confidentiality level. |
dataset-dependency-relationship | Dataset dependency relationship | BOM-G7AI-DP-009 G7AI-DP-09 | Requirement | Relationships linking a model to the datasets it was trained on, tested on, or depends on. |
dataset-license | Dataset license | BOM-G7AI-DP-010 G7AI-DP-10 | Requirement | The declared or concluded license under which the dataset is distributed, including access and usage restrictions. |
infrastructure | Infrastructure | G7AI-INF | Elements describing the software and hardware environment on which the AI system runs. | |
infrastructure-software | Infrastructure software | BOM-G7AI-INF-001 G7AI-INF-01 | Requirement | The software environment required to run the AI system, including container images, OS, ML frameworks and runtime libraries. |
infrastructure-hardware | Infrastructure hardware | BOM-G7AI-INF-002 G7AI-INF-02 | Requirement | The hardware required to operate the AI system, including specialised accelerators (GPUs, TPUs, NPUs). |
key-perf-indicators | Key Perf Indicators | G7AI-KPI | Elements describing measurable security and operational perf metrics for the AI system. | |
security-metric | Security metric | BOM-G7AI-KPI-001 G7AI-KPI-01 | Requirement | Quantitative security-related perf metrics and their decision thresholds (e.g. adversarial robustness scores, CVSS scores). |
operational-perf-kpi | Operational perf KPI | BOM-G7AI-KPI-002 G7AI-KPI-02 | Requirement | Operational performance indicators for the AI system such as accuracy, F1, latency, throughput and energy consumption. |
metadata | Metadata | G7AI-MD | Elements describing the document itself: authorship, format, versioning, tooling and relationships. | |
sbom-author | Document author | BOM-G7AI-MD-001 G7AI-MD-01 | Requirement | The agent (person, organisation or tool) that produced the document. |
sbom-version | Document version | BOM-G7AI-MD-002 G7AI-MD-02 | Requirement | A version identifier for this specific instance of the document. |
sbom-data-format-name | Data format name | BOM-G7AI-MD-003 G7AI-MD-03 | Requirement | The name of the standard or schema used to serialise the SBOM (e.g. SPDX, CycloneDX). |
sbom-data-format-version | Data format version | BOM-G7AI-MD-004 G7AI-MD-04 | Requirement | The version of the SBOM data format specification used (e.g. 3.1, 1.6). |
sbom-author-signature | Author signature | BOM-G7AI-MD-005 G7AI-MD-05 | A cryptographic signature or integrity method that allows verification of SBOM authorship. | |
sbom-tool-name | Tool name | BOM-G7AI-MD-006 G7AI-MD-06 | Requirement | The name of the software tool used to generate the SBOM. |
sbom-tool-version | Tool version | BOM-G7AI-MD-007 G7AI-MD-07 | Requirement | The version of the software tool used to generate the SBOM. |
sbom-generation-context | SBOM generation context | BOM-G7AI-MD-008 G7AI-MD-08 | Requirement | The lifecycle stage or context in which the SBOM was generated (e.g. build, release, deploy, runtime). |
sbom-timestamp | Document timestamp | BOM-G7AI-MD-009 G7AI-MD-09 | Requirement | The date and time at which the SBOM was created. |
sbom-dependency-relationship | Document dependency relationship | BOM-G7AI-MD-010 G7AI-MD-10 | Explicit dependency relationships between components recorded in the SBOM. | |
models | Models | G7AI-MDL | Elements describing individual AI model artifacts: identity, provenance, hashes, licensing and training characteristics. | |
model-name | Model name | BOM-G7AI-MDL-001 G7AI-MDL-01 | Requirement | The human-readable name of an individual AI model component. |
model-identifier | Model identifier | BOM-G7AI-MDL-002 G7AI-MDL-02 | Requirement | A stable, unique identifier for the model (e.g. PURL, SPDX ID, Hugging Face model ID, DOI). |
model-version | Model version | BOM-G7AI-MDL-003 G7AI-MDL-03 | Requirement | A version or checkpoint identifier for the specific weight snapshot of the model. |
model-timestamp | Model timestamp | BOM-G7AI-MDL-004 G7AI-MDL-04 | Requirement | The date and time the model was trained, built, or released. |
model-producer | Model producer | BOM-G7AI-MDL-005 G7AI-MDL-05 | The organisation or person that trained or supplied the model. | |
model-description | Model description | BOM-G7AI-MDL-006 G7AI-MDL-06 | Requirement | A free-text description of the model's purpose, architecture and key characteristics. |
model-hash-value | Model hash value | BOM-G7AI-MDL-007 G7AI-MDL-07 | The cryptographic digest of the model weight file(s) for integrity verification. | |
model-hash-algorithm | Model hash algorithm | BOM-G7AI-MDL-008 G7AI-MDL-08 | The algorithm used to compute the model hash (e.g. SHA-256, SHA-512, BLAKE3). | |
model-prop | Model property | BOM-G7AI-MDL-009 G7AI-MDL-09 | AI-specific technical properties of the model such as type, explainability, autonomy level and energy profile. | |
model-input-output-prop | Model input/output property | BOM-G7AI-MDL-010 G7AI-MDL-10 | The types, modalities and constraints on the model's input and output (e.g. image, text, structured data). | |
model-training-prop | Model training property | BOM-G7AI-MDL-011 G7AI-MDL-11 | Characteristics of the model's training process, including hyperparameters, data pre-processing and fine-tuning steps. | |
model-license | Model license | BOM-G7AI-MDL-012 G7AI-MDL-12 | Requirement | The declared or concluded license under which the model is distributed (e.g. SPDX expression, AI-specific RAIL license). |
model-external-ref | Model external reference | BOM-G7AI-MDL-013 G7AI-MDL-13 | Requirement | Links to supplementary resources such as model cards, research papers, repositories and model registries. |
system-level-prop | System Level Properties | G7AI-SLP | Elements describing the AI system as a whole: name, components, producer, version, timestamps, data flows and intended use. | |
system-name | System name | BOM-G7AI-SLP-001 G7AI-SLP-01 | Requirement | The name of the AI system as a whole. |
system-components | System components | BOM-G7AI-SLP-002 G7AI-SLP-02 | The constituent sub-components (models, datasets, libraries) that make up the AI system. | |
system-producer | System producer | BOM-G7AI-SLP-003 G7AI-SLP-03 | The organisation or person that developed or supplied the AI system. | |
system-version | System version | BOM-G7AI-SLP-004 G7AI-SLP-04 | Requirement | The release or version identifier of the AI system. |
system-timestamp | System timestamp | BOM-G7AI-SLP-005 G7AI-SLP-05 | Requirement | The date and time the AI system was built or released. |
system-data-flow | System data flow | BOM-G7AI-SLP-006 G7AI-SLP-06 | Requirement | A description of how data flows into, through and out of the AI system. |
system-data-usage | System data usage | BOM-G7AI-SLP-007 G7AI-SLP-07 | A description of how data (including personal or sensitive data) is used by the AI system. | |
system-input-output-prop | System input/output property | BOM-G7AI-SLP-008 G7AI-SLP-08 | Requirement | The types, modalities and constraints of data accepted as input and produced as output by the AI system. |
system-intended-application-area | Intended application area | BOM-G7AI-SLP-009 G7AI-SLP-09 | The domain(s) or application area(s) for which the AI system is intended (e.g. healthcare, finance, autonomous driving). | |
security-prop | Security Properties | G7AI-SP | Elements describing security controls, compliance, policies and known vulnerabilities. | |
security-controls | Security controls | BOM-G7AI-SP-001 G7AI-SP-01 | Requirement | The security controls implemented on the AI system (e.g. access controls, encryption, audit logging). |
security-compliance | Security compliance | BOM-G7AI-SP-002 G7AI-SP-02 | Requirement | The security standards, regulations or frameworks the AI system has been assessed against or conforms to. |
cybersecurity-policy | Cybersecurity policy information | BOM-G7AI-SP-003 G7AI-SP-03 | Requirement | References to the supplier's vulnerability-disclosure policy, security.txt or coordinated-disclosure process. |
vulnerability-referencing | Vulnerability referencing | BOM-G7AI-SP-004 G7AI-SP-04 | Requirement | References to known vulnerabilities (CVEs) and their assessment status (VEX) for the AI system and its components. |