Start anywhere
Each page answers in the first sentence, states what the base framework establishes, shows whether the public draft touches it, lists the evidence a reviewer samples, names the common gaps and gives a preparation step. The base sample is held for clauses 1 to 6.3. Later clauses are described from the public contents and public summaries, and the page says so.
The amendment body is not held. No page quotes it or invents it. Where the public draft summary names an addition, the page says it is a summary and labels the final wording as not held.
Clause 1 and references
The scope of the framework and the one document it depends on for vocabulary.
| Unit | Page title | Amendment status |
|---|---|---|
1 | Scope | Not touched by the draft |
2 | Normative references | Not touched by the draft |
Terms and definitions
The words a model card and a data statement need in order to be readable.
| Unit | Page title | Amendment status |
|---|---|---|
3.1 | Model development and use | Not touched by the draft |
3.2 | Tools | Not touched by the draft |
3.3 | Data | Not touched by the draft |
Abbreviated terms
The short forms used through the base text.
| Unit | Page title | Amendment status |
|---|---|---|
4 | Abbreviated terms | Not touched by the draft |
Overview
The map of an ML system inside the wider AI ecosystem.
| Unit | Page title | Amendment status |
|---|---|---|
5 | Overview | Not touched by the draft |
Clause 6: Machine learning system
Task, model, data and tools, with the algorithm catalogue where the draft adds its two new entries.
| Unit | Page title | Amendment status |
|---|---|---|
6.1 | Overview of the machine learning system | Not touched by the draft |
6.2 | Task | Not touched by the draft |
6.3 | Model | Not touched by the draft |
6.4 | Data | Not touched by the draft |
6.5 | Tools | New material added by the draft |
6.5.3 | Categories of machine learning algorithms | New entries added by the draft |
Amendment 1: new generative AI material
What the public draft adds, and what a governance record has to do about it.
| Unit | Page title | Amendment status |
|---|---|---|
6.5.3.2.9 | Transformer architecture and algorithm | New in the draft |
6.5.3.2.10 | Generative model | New in the draft |
BIB | Bibliography and the AI-generated content reference | New reference in the draft |
Clause 7: Machine learning approaches
Supervised, unsupervised, semi-supervised, self-supervised, reinforcement and transfer learning.
| Unit | Page title | Amendment status |
|---|---|---|
7.1 | General | Not touched by the draft |
7.2 | Supervised machine learning | Not touched by the draft |
7.3 | Unsupervised machine learning | Not touched by the draft |
7.4 | Semi-supervised machine learning | Not touched by the draft |
7.5 | Self-supervised machine learning | Note expanded in the draft |
7.6 | Reinforcement machine learning | Not touched by the draft |
7.7 | Transfer learning | Not touched by the draft |
Clause 8: Machine learning pipeline
Data acquisition, preparation, modelling, verification, deployment and operation.
| Unit | Page title | Amendment status |
|---|---|---|
8.1 | General | Not touched by the draft |
8.2 | Data acquisition | Not touched by the draft |
8.3 | Data preparation | Not touched by the draft |
8.4 | Modelling | Not touched by the draft |
8.5 | Verification and validation | Not touched by the draft |
8.6 | Model deployment | Not touched by the draft |
8.7 | Operation | Not touched by the draft |
8.8 | Example machine learning process based on the pipeline | Not touched by the draft |
Annex A (informative)
The example data flow and data use statement.
| Unit | Page title | Amendment status |
|---|---|---|
A | Annex A: Example data flow and data use statements | Not touched by the draft |