Pick the kind of data above and the result will appear here.
This is our practical starting point. Your organisation's own rules always win.
The whole picture
The same guidance as a plain table. On a small screen, scroll the table sideways.
| Kind of data | Public AI tools | Your employer's approved AI tools | A build in your organisation's own environment |
|---|
Three traps
1. "No names" is not anonymous
A level, a job family and a pay figure can point to one person in a small team. If a colleague could guess who it is, treat it as identifiable.
2. Available is not approved
A tool being available at work is not the same as being approved for pay data. Approval is usually for a stated kind of data, so check which kinds yours covers.
3. In your browser or sent away
Some tools send what you type to a server run by someone else. Others do their work on your own computer and send nothing. A tool like our pay gap analyser does its work on your own computer.
Five questions to ask about any AI tool
Terms differ by product and by plan, and they change. Ask these of the tool you are actually using, and look for the answer in your organisation's agreement.
- Where does the data go and where is it stored?
- Is it used to train or improve the provider's models?
- Who at the provider or in our organisation can see it?
- How long is it kept and can we delete it?
- Has our organisation approved it for this kind of data?
Still unsure?
- Use invented data until you are sure.
- Ask your privacy officer or your IT security team.
- Read our short primer on the Australian Privacy Principles.
General information, not legal advice. Your organisation's own rules always win.