top of page

AI Use and Disclosure Policy

At a glance

The Proptech Advocate uses artificial intelligence to help find, research and draft stories. A human editor checks and approves everything before it is published, and an editor is accountable for every word on this site.


  • We tell you when AI was involved. Every article carries a note explaining what AI did and who reviewed it.
  • AI never invents. We do not use AI to create quotes, sources, people, data or realistic images of real events.
  • Humans decide. Editors decide what we cover, what we say and what we publish. AI does not.
  • Money does not buy coverage. AI is held to the same rule as our journalists: paid content is always labelled, and advertisers do not shape our news.
  • Tell us if we get it wrong. Errors are corrected openly, whether a person or a machine made them.


1. Purpose and scope

This policy explains how The Proptech Advocate ("The Advocate", "we") uses artificial intelligence in its journalism, and how we tell readers about it.


It applies to everything published under The Advocate masthead: the website, newsletters, podcasts and video, and our social media channels. It covers our staff, contributors, columnists and technology partners, and any AI tool used to produce or distribute our content.


In this policy, AI means software that generates or changes text, images, audio or video, or that analyses information to make suggestions. It includes generative AI tools such as large language models and image generators.


This policy sits alongside our Editorial Standards and Privacy Policy. Where they overlap, the stricter rule applies.


2. Our principles

Five principles govern every use of AI at The Advocate. They follow the Australian Press Council's Statements of Principles and Standards of Practice, which apply in full whether or not AI was used.


  1. Transparency. We do not hide our use of AI. We explain what it did, in plain language, where readers will see it.
  2. Human accountability. A named editor is responsible for everything we publish. "The AI did it" is never an excuse.
  3. Accuracy first. AI can produce confident errors. We treat AI output as an unverified source and check it before it is published.
  4. Editorial independence. AI is a tool for journalism, not for selling. It does not decide coverage, and commercial partners have no access to or influence over our editorial AI tools.
  5. Respect for others' work. We use AI in ways that respect copyright, credit original reporting and protect personal information.


3. How we use AI

AI supports our journalists at most stages of production, from finding stories to sharing them. At every stage, a person makes the decision.


Finding stories

  • What AI does: Scans public news, announcements and industry sources each week and suggests possible stories
  • What a person does: Editors choose which stories to pursue, if any


Research

  • What AI does: Summarises long documents such as reports, filings and transcripts, and finds relevant passages
  • What a person does: Journalists read the original source for anything we rely on


Drafting

  • What AI does: Produces a first draft, or rewrites a media release into our house style
  • What a person does: An editor rewrites, checks facts and approves the final text


Editing

  • What AI does: Suggests headlines, checks spelling, grammar and style, and flags possible errors
  • What a person does: An editor accepts or rejects every suggestion


Search and discovery

  • What AI does: Suggests keywords, summaries and metadata so readers can find our work
  • What a person does: An editor confirms these accurately reflect the story


Distribution

  • What AI does: Prepares and schedules social media posts and newsletter summaries
  • What a person does: Posts follow pre-approved rules, and an editor reviews them regularly


Audio and video

  • What AI does: May be used for transcription, captions and editing
  • What a person does: Hosts and producers review the output; AI is not used to imitate a real person's voice or likeness


We may add new uses as tools improve. When we do, we will update this policy first.


4. What AI is never used for

Some uses are off limits, whatever the deadline or the commercial pressure. We will never use AI to:

  • invent or alter quotes, or attribute words to anyone who did not say them
  • create sources, interviewees, experts or people who do not exist
  • produce statistics, data or market figures that do not come from a verified source
  • create realistic images, audio or video of real people, places or events that could mislead readers about what happened
  • imitate the voice, likeness or writing style of a real, identifiable person without their consent
  • publish content that no human has reviewed
  • decide what is newsworthy, or decide the angle or conclusion of a story
  • write opinion pieces, columns or editorials published under a person's name, beyond spelling, grammar and style suggestions
  • hide the fact that a story is sponsored or paid for


5. Human oversight and accountability

No article is published on The Advocate until a human editor has reviewed and approved it. The Editor is accountable for all published content, including any errors introduced by AI.


Before approving AI-assisted work, the reviewing editor must:

  1. check every fact, figure, name, date and quote against an original source
  2. confirm that every quote was actually said or written by the person it is attributed to
  3. confirm that sources are credited and that the story is not a close copy of another publisher's work
  4. check that the story is fair and balanced, and that the headline matches it
  5. make sure the correct AI disclosure is attached (see section 6)


The people who write the rules that guide our AI tools are accountable for those rules. Those instructions are reviewed against this policy and our Editorial Standards before they are used, and whenever they change.


We keep a record of which AI tools were used on each story and who approved it, so we can answer questions and investigate errors.


6. How we label AI involvement

Every article carries a short disclosure note at the end, matched to how much AI contributed. Readers should know what AI did and who checked it, without having to look for it.


Bylines. Stories written mainly by a journalist carry that journalist's name. Stories produced with substantial help from AI and edited by our team carry the byline Advocate Staff. We never give AI-produced work the name of a real or invented person.


Disclosure notes. We use three levels of disclosure:


Human-written

  • When it applies: A journalist researched and wrote the story. AI was used only for spelling, grammar, transcription or search suggestions.
  • Note shown at the end of the article: This article was written by [name]. Learn how we use AI [link].


AI-assisted

  • When it applies: AI helped research, summarise or draft the story. A journalist or editor substantially rewrote, fact-checked and approved it.
  • Note shown at the end of the article: This article was produced with the help of AI tools and was checked and edited by [editor name] before publication. Learn how we use AI [link].


AI-generated

  • When it applies: Most of the text was produced by AI, such as a summary of a report or a market data update. An editor fact-checked and approved it with limited rewriting.
  • Note shown at the end of the article: This article was generated with AI from [source] and was fact-checked and approved by [editor name] before publication. Learn how we use AI [link].


Placement. For AI-generated articles, a short label (AI-generated, human-checked) also appears at the top of the story, near the byline. Labels stay with the content when it is shared on social media and in newsletters.


Images and audio. These carry their own label beside them, not only a note at the end of the article (see section 7).


This tiered approach follows the Australian Government's voluntary guidance Being clear about AI-generated content, which recommends matching transparency to the level of AI involvement and the impact of the content.


7. Images and visual content

We prefer real photographs from licensed libraries, our partners or our own photographers. AI-generated images are used only for illustration, never as evidence of what happened.


  • Labelled every time. Any AI-generated or significantly AI-altered image carries a caption such as Illustration: AI-generated. This follows the Press Council's long-standing expectation that digitally altered images are disclosed.
  • No fake reality. We do not create realistic images of real people, real properties or real events.
  • Minor edits are fine. Routine adjustments such as cropping, colour correction and brightness are not labelled, provided they do not change what the image shows.
  • Charts and data visuals. Where AI helps build a chart, the underlying data comes from a named, verified source, which we credit.


8. Sponsored, partner and contributed content

Paid content is always clearly labelled as paid, whether or not AI helped make it. The AI disclosure rules in this policy apply to it as well.


  • Sponsored and partner content carries a prominent Sponsored or Partner content label at the top. If we used AI to help produce it, the AI disclosure note also appears.
  • Editorial AI stays editorial. The AI tools and instructions we use to find and write news are not used to promote partners, and partners cannot ask them to cover or avoid a topic.
  • Contributors and guest columnists must tell us if they used AI to write their piece. They remain responsible for its accuracy and originality, and we label it according to section 6.
  • Media releases are never republished word for word. If AI rewrites a release, an editor checks it against the original and the story is labelled AI-assisted.


9. Accuracy, corrections and complaints

If we publish an error, we correct it promptly and say so, whether a person or AI caused it. We do not quietly edit published stories to hide mistakes.


  • Corrections are noted at the end of the article, with the date and what changed.
  • If an error came from AI output that we failed to catch, we review the AI tool's instructions and our checking process, and fix the cause.
  • Readers can report errors or concerns about our use of AI to [editor email]. We aim to respond within [two business days].
  • Full details are in our Editorial Standards.


10. Data, privacy and the AI tools we use

We protect personal and confidential information when using AI. We use business versions of AI tools whose providers agree not to train their models on the information we give them.


  • We do not put readers' personal information, subscriber lists or private correspondence into AI tools, except where our Privacy Policy allows it and the tool meets our security requirements.
  • We do not put confidential source material or unpublished information from interviewees into AI tools without the source's agreement.
  • We do not use AI to make decisions that significantly affect individuals, such as decisions about subscriptions or access to our services, without human review. If AI helps moderate reader comments, a person reviews any appeal.
  • Our Privacy Policy explains how we handle personal information, including any automated decision-making, in line with the Australian Privacy Principles.


On request, we will tell readers which categories of AI tools we use, such as large language models and image generators.


11. Copyright and respect for original journalism

We credit and link to the original source of any story we follow up, and we do not use AI to repackage another publisher's reporting as our own.


  • When AI helps us find a story reported elsewhere, we credit the outlet that broke it and add our own reporting, context or analysis.
  • AI-assisted drafts are checked to make sure they do not copy substantial passages from other publications.
  • We only use AI to process paywalled or licensed material where our licence allows it.
  • We use images and other media only where we hold the rights or a licence to use them.


12. Review and contact

We review this policy at least every six months, and sooner if our tools, the law or industry standards change. The next review is due by 13 May 2027.


We will update this page before introducing any significant new use of AI, and record the date of each change below.


Questions or concerns? Contact the Editor at [editor email].


Policy effective: [launch date]. Last updated: [date].


Sources and references

bottom of page