Data is a component in its own right, not a by-product of the model. The clause covers the kinds of data an ML system uses and the care they need.
What the base framework establishes
The public contents gives this clause one page and no titled subclauses. The framework treats data as one of the three components of the system, alongside model and tools, and the surrounding clauses show what that means in practice: data is acquired for a purpose, prepared and split, labelled where the task needs it, and protected where it identifies people. The clause sits between the model and the tools so that a reader cannot describe a system without saying what it learned from. The public sample does not carry the clause text, so this page states the component's place in the framework and does not put words into the clause.
Why it matters to an AI management system
Treating data as a component makes it assignable. A named data owner can be asked for the extract, the licence, the de-identification and the split. Without that, every data question becomes an engineering favour.
The public draft summary names no change to 6.4 Data. The final text could still touch it.
Evidence a reviewer samples
| Category | Artefact |
|---|---|
| Data inventory | Each dataset, its source, licence, extract date and owner. |
| Split record | How training, validation and test sets were separated and why they do not overlap. |
| Privacy note | The de-identification method and the personal data categories removed. |
One question to prepare for. Who owns the training data, and what licence covers its use?
Common gaps
- The dataset has no named owner and no licence record.
- The same rows appear in training and test sets, inflating the result.
- Personal data is described as anonymised without a method.
Prepare before publication
Give every dataset an owner and a licence line. Record the split method and check for overlap before the next evaluation.
Dates that matter
No separate date attaches to 6.4; the record dates are the stage date, the planning estimate and the CEN expectation stated above.
Cross-references
3.3 Data · 8.2 Data acquisition · 8.3 Data preparation
Source
Source for 6.4: ISO/IEC 23053:2022, clause 6.4. The public sample stops before 6.4, so this page relies on the public contents and the public summary and does not quote the clause. The two official records are linked on this page so any line can be checked.
Not legal advice; does not certify. A reading of the framework and the evidence a reviewer commonly samples, not a substitute for the standard or your auditor's rules.