13 March 2026
Joseph Keriwyn
(no address supplied)
Email: [FYI request #34004 email]
Tēnā koe Joseph,
Thank you for your OIA request regarding this article
: Ōtara parents question pool ‘manu’
restrictions ahead of World Championships | RNZ News. In answer to your questions:
- The paragraph that concerns you was not written by a large language model (AI).
RNZ does not use such technology in preparing its stories. In this particular case,
the paragraph was added by one of the editing team, which is why the tone is
different from that of the main article.
- RNZ has not received any complaints about this article.
- A copy of RNZ’s AI principles is included in this response.
- All newsrooms employing an LDR reporter are obliged by contract to ensure all
copy meets RNZ’s editorial standards.
- This particular article has had 10,238 pageviews as at 10am, 13 March 2026.
I trust this answers your questions and reassures you that AI is not utilised in the
preparation of news articles that appear on RNZ’s website.
You have a right to seek an investigation and review by the Ombudsman of the above
outcome. Information about how to make a complaint is available at
www.ombudsman.parliament.nz or freephone 0800 802 602.
A referral must occur within 20 working days.
Ngā mihi,
Andrew Holden (he/him)
OIA Officer
PO BOX 123 Wellington, New Zealand +64 4 474 1999 www.rnz.co.nz
RNZ Artificial
Intel igence Principles
-free public media organisation. Our vision is to create
Outstanding Public Media that Matters. New technological tools offer huge opportunities to
modernise and change the way we work and tel stories, but we also need to be mindful of
our core role: to serve the public interest.
Whenever we try new things at RNZ, it is important we do so in a way that upholds trust and
is consistent with RNZ's Charter. Our trust and credibility are taonga we must protect.
Like many technological advancements, used appropriately, AI may be a powerful tool, but
it is not a solution in and of itself. It may deliver new efficiencies and innovations and al ow
us to deliver more, and be more accessible,
also brings risks
around the law, copyright, ethics, and mis and disinformation. Above al , RNZ's audiences
need to be able to trust that our storytel ing is robust, credible, and transparent.
Definitions
Artificial intel igence is general y thought of as a machine-based system that can perform
tasks usual y requiring human intel igence, such as decision-making, translation, and/or
speech recognition.
Machine Learning is one type of AI, where computer systems can learn from data patterns
using algorithms and statistical analysis, often without any intervention from a human.
These might sometimes be described as assistive.
Generative AI is another branch of AI that can create content, including images, speech,
text, music, video and other data fol owing requests from a user.
Principles
With the above in
six principles to guide
work (and that done
on our behalf) on the use of AI.
RNZ wil general y not publish, broadcast or otherwise knowingly disseminate work created
by generative AI. However, we acknowledge that it may not always be possible to know the
provenance of work created by one of our many partners or suppliers, and that as
technology changes and adapts, so may our policies.
Any use of AI, generative or otherwise, should be done in consultation with senior
managers. When in doubt, always refer up. The AI working group wil oversee al AI work1.
Oversight
Any use of generative AI for any kind of content - should be for a specific purpose that
serves audiences (e.g explaining how it has been used by other services). It should always be
with the explicit permission of a senior editor.
As with al RNZ content, before anything is published or broadcast, content created by
generative AI should be checked by a human. We should never assume that the information
used by AI services is accurate. The human check should include the editorial principles of
independence, accuracy, and fairness, and any content must be in line with the editorial and
social media policies.
The use of AI outside journalism and storytel ing also needs to be approved by a senior
manager and should serve a specific purpose. We should always endeavour to know what
and how the AI is used.
The success of any AI is based on the algorithm and data it is trained on. It is very easy for
inherent biases to be imposed on that data. Any use of any AI for RNZ must consider
whether the appropriate checks have taken place to measure and mitigate those biases.
Transparency
We are open with our audience about how we have created stories and what tools we used.
and process apply. Disclosures should be precise in language without being confusing to our
audience. Human oversight wil always be a key factor in our publishing and broadcasting.
Internal y, AI processes should be documented and described so we can communicate,
debate and learn from what has been used.
If we're contracting a service or product, we should make sure we know what AI is used.
Policies and the law
Any AI policy or projects wil need to include the application of the principles of Te Tiriti, the
RNZ Charter and the RNZ strategy.
Policies and projects should also make sure that no laws e.g. copyright are being broken
in the use of AI for RNZ. As we utilize AI to customize content or develop products that alter
privacy policy.
Where we are using services or content provided by third parties, we should make every
endeavour to understand what role AI has played in their creation.
1 Current membership:
Stephen Smith; Troy Stanton; Mark
Jamie Annan; Brendon
Murphy; Rhonwyn Newson
Serving the public
Use of AI can transform how RNZ works and may in the future make us more efficient and
better able to deliver value to the public. Any use of AI should serve our purpose: to
strengthen Aotearoa and the Pacific through informative, chal enging and entertaining
content.
That needs to be balanced against the chal enges it brings, including trust in the media,
Prioritising talent and creativity
We work with the best people in their field, and we wil continue to do so AI cannot
replace the brain of a reporter, storytel er, producer, technician, or broadcaster doing what
, both in its use of AI,
and in the way generative AI creates its content.
Security and Privacy
Ensuring security and safety of RNZ systems, free from malicious use and cyber threats must
be of paramount importance in any AI policy or project.
AI systems can use private data in unethical or unlawful ways or with little to no
transparency. Protecting the privacy and personal data of individuals and ensuring
compliance with law, wil always be of utmost importance to RNZ.
Editorial-specific guidance
As always, when in doubt, refer up.
Generative AI should not be used to create news stories, current affairs or factual
journalism for RNZ. The exception to this rule is when AI is the subject of the story,
and its use is to il ustrate that, or to educate our audience.
Other exceptions may be made on a case-by-case basis subject to approval and
guided by these principles.
As always, our journalism is guided by the RNZ editorial policy. AI-generated material
should be treated as unverified source material, which needs to be vetted and
confirmed by a human. If there are any doubts. AI models are learning from data
from the past they may not be ful y updated. They can also make things up this is
commonly cal ed hal ucinating so as ever, anything we might consider using,
whether it is the summary of a report, suggested headline, or col ection of data
needs to be careful y verified.
If AI is used as the subject of a story (e.g., an image created), it should be clearly
marked as such and explained. If an image or other work has inadvertently been
used, it should be corrected, and a correction clearly issued.
Photos and video should not be manipulated beyond cropping and editing for colour,
contrast, and brightness.
RNZ kaimahi should not input confidential or sensitive information into AI tools.
If journalists or editors do not have confidence in the authenticity of the material,
whether it is text, audio, or images, they should not use it.
Last updated: 15/8/24
