AI Disclosure

Where we use it

• Drafting and editing text from sources that are being verified.
• Translation across the nine editions.
• Suggestion and triage: ordering topics, summaries and candidate headlines.
• Editorial illustrations, labelled as illustrations.

Where we refuse

We do not generate images that imitate press photography. A realistic-looking image of an event that did not happen — or that we did not witness — is precisely what we ask others not to do, and a small disclosure line beneath it does not change that.

We do not generate a quotation attributed to a real person. A quote is either recorded, documented, or non-existent.

We do not generate a source, a reference or a figure. What we cannot trace to something published does not enter the text.

How we label images

• A real photograph — carries its photographer or source and its licence.
• Editorial artwork — Voice of the Citizen: visual work of our own that makes no claim to document anything.
• Generated illustration, not a photograph: generated artwork explaining an abstract idea that cannot be photographed.
• Citizen Lens: a reader’s photograph, credited to them after their rights declaration and our review.

Responsibility

Editorial supervision is human, and responsibility for what is published is entirely human. "The system got it wrong" is not an excuse we accept, and it is not one we will use.

Every piece passes through a documented editorial workflow before publication, and every intervention is recorded against the person who made it.

Reader privacy

We use no third-party trackers or analytics, and we feed no reader data into any system. Our readership measurement is aggregate: we know a piece was read; we build no profile of who read it.

This policy will change

The tools change faster than the norms. When our practice changes we change this text first — not after somebody discovers it.