AI decision log, model cards and privacy statement - Product News
Every AI output now has a decision-log entry naming who acted on it, each AI capability has a published model card, and Backed writes the automated-decision statement for your privacy policy.
Published
ForAdministrators
What's new
- The decision log. A new
Decision logtab on AI assistance lists every AI output: the capability, the record it was for, the model and prompt versions, and who accepted, edited, noted or discarded it, and when. An output nobody acts on within 30 days is markedNot used. The log keeps references and hashes, never the prompt or the output, and no one can edit or delete an entry. Filter it by capability, record, person or date, and selectExport CSVto download what you filtered. - AI assisted in record histories. When an accepted AI output changes a record, its history
shows an
AI assistedline that opens the decision log entry. - Published model cards. Each AI capability has a model card saying what it is for, the model
and provider, where requests are processed, what it uses and never receives, what it never does,
where a person approves, its limits and when it was last evaluated. The card is shown when you
switch a capability on, and on a public
AI model cardspage on your applicant portal that anyone can read without signing in. - The privacy statement. A new
Privacy statementtab writes the automated-decision statement for your privacy policy from the capabilities you have on: the kinds of personal information they use, the decisions each one helps with, and that no decision is made by a computer program alone. It is dated, and it is written again whenever a capability changes. SelectCopy statementto paste it into your policy.
Why it matters
From 10 December 2026 the Privacy Act requires an organisation to say in its privacy policy when a computer program helps make decisions about people. The statement, the model cards and the decision log let you say what AI does in your grant-making, show that a person decides, and answer for every output afterwards. Have your own adviser read the statement before you publish it.
Where to find it
Settings > Profile and terms, on the AI assistance tab: Decision log and Privacy statement
sit beside Settings and Usage.
Was this helpful?