AI in Creative Industries

Executive Summary

The previous update’s central conclusion still holds: AI adoption is broad but uneven, and the strongest evidence concerns assistive workflows rather than autonomous public-facing production.

Since 1 September 2026, the evidence base has improved in three ways:

New Zealand journalism now has a larger documented set of operating deployments, including transcription, translation, research, metadata and content transformation.
Music and publishing organisations have moved from general concern to explicit positions on licensing, transparency, Māori cultural property and creator control.
Government and industry support for small-business adoption has expanded, although there is still no public breakdown showing how many creative businesses are participating or what systems they have deployed.

The main development is therefore not a sudden shift to AI-generated films, books, music or visual art. It is the institutionalisation of AI around creative work: organisations are deciding which tools can be used, what material can be supplied, where human review is required and how use should be disclosed.

The evidence remains strongest in journalism and publishing, followed by marketing and selected language or accessibility applications. Screen, games, music production, visual arts and independent creative practice show substantial experimentation and policy activity, but public evidence of scaled deployment remains limited.

A small qualitative study from Massey University also sharpens the workforce picture. Its ten interviewees could not identify specific examples of paid creative roles already being displaced by generative AI, but they did report substitution of low-paid, freelance and unpaid work that traditionally helps new entrants build experience and networks. The study is small and exploratory, but its warning is more specific than general claims about automation.

The emerging Aotearoa model remains human-led, workflow-oriented and governed through rights policy, public funding, cultural consultation and professional norms.

What Has Changed Since the Last Update

Music rights have become a more detailed election issue

On 16 September 2026, Recorded Music NZ published AI, Copyright and the Future of NZ Music. The paper follows the broader music-industry manifesto published in August but makes the requested policy settings more operational.

Recorded Music NZ is calling for:

No weakening of existing copyright protection to facilitate AI training.
AI companies to license music through ordinary rights-holder channels.
Meaningful records of the material used to train AI systems.
A coordinated Government response covering copyright, employment, cultural effects and wider AI risks.

This is a sector position, not current law. Its significance is that the music industry is no longer arguing only against “AI” in general. It is distinguishing between creator-controlled use of assistive tools and commercial systems trained on music without permission or transparent records. nzmusic.org.nz

The issue is now part of the 2026 election environment. Labour’s AI Action Plan proposes a copyright framework under which creators control and are paid for the use of their work. That is also a policy proposal rather than enacted law. labour.org.nz

Publishing has formalised its position

The Publishers Association of New Zealand – Te Rau o Tākupu has published an Artificial Intelligence Position Statement. It recognises existing use of AI across publishing operations and creative processes, including potential benefits for accessibility, discoverability, efficiency and innovation.

The statement also places publishing’s AI position within a specifically Aotearoa framework:

Existing New Zealand copyright law continues to apply.
Human creativity and professional contribution should be respected.
AI development and use should be transparent and accountable.
Māori language, knowledge and creative works require careful consideration and engagement consistent with Te Tiriti o Waitangi.
Innovation should support, rather than undermine, the sustainability of publishing and authorship.

This is evidence of sector governance and advocacy, not evidence that publishers have adopted AI at a particular scale. The statement is nevertheless important because it shows that publishing is moving toward organisational principles rather than leaving AI decisions entirely to individual authors or suppliers. publishers.org.nz

Small-business support has expanded, but creative adoption is still unmeasured

On 22 September 2026, the Government extended its AI Advisory Pilot to support a further 120 small businesses through to 30 June 2027.

The pilot provides:

Expert advice on suitable AI use cases.
A tailored AI roadmap.
Co-funding of implementation costs of up to $15,000.
Support delivered through the Regional Business Partner Network.

The Government says around 150 businesses had already joined the pilot, while approximately 650 businesses attended AI-related courses funded through the network between July 2025 and June 2026. These figures cover the wider small-business population, not the creative industries specifically. There is no public evidence yet showing how many creative agencies, studios, publishers, artists or production companies have received support, or whether participation has resulted in operating deployments. beehive.govt.nz

The extension is still relevant to creative industries because much of the sector consists of small companies, sole traders and project-based businesses that lack dedicated technology staff. It reduces an adoption barrier, but it does not by itself demonstrate adoption.

Capability-building is becoming more visible than deployment

Two September initiatives illustrate the shift toward capability and governance.

On 23 September, the University of Auckland launched Huanui: The AI Initiative. The initiative is intended to support research, teaching, business engagement and responsible AI practice, with mātauranga Māori and Indigenous perspectives among its stated foundations. It is a capability and research hub, not a creative-industry deployment programme. auckland.ac.nz

The AI Forum’s Creative Industries workstream also held a public update on 22 September. Its wider Blueprint calls for baseline data on AI use by New Zealand creative professionals and businesses, clearer distinctions between operational AI and generative creative output, practical training and local examples of responsible implementation. aiforum.org.nz

The implication is that the sector is still building the conditions for adoption. Capability programmes, principles and research partnerships are progressing faster than publicly documented commercial deployments.

The proposed national creative AI platform is not yet evidence of operation

The Aotearoa Creative AI Research Institute, led by Wētā FX, remains documented as one of five concepts shortlisted to develop proposals for the national AI research platform. The concept includes computer vision, imaging, complex data models, digital twins and AI rights management.

The New Zealand Institute for Advanced Technology formally launched in August 2026 and says it is progressing work in AI. However, public information reviewed for this update does not establish that the Aotearoa Creative AI Research Institute has been selected as the final funded platform or that it is operating as a functioning research institute. The concept should therefore be treated as a research proposal within a national funding process, not as an active sector-wide capability. mbie.govt.nz

The core evidence gaps remain

The Survey of Business Operations 2026 includes questions on business AI adoption, impacts, workforce and innovation. Results are still not publicly available.

This means New Zealand still lacks a reliable national measure of:

AI adoption by creative subsector.
The number of businesses using AI in production or administration.
Changes in staffing, freelance work or revenue.
Productivity gains attributable to AI.
The proportion of AI-generated material reaching audiences.
Differences between large organisations, small firms, independent practitioners and Māori or Pacific-led organisations.

The previous update’s caution about the evidence base therefore remains justified. mbie.govt.nz

Current State of AI Adoption

Journalism and publishing: the clearest operating evidence

Journalism remains the most documented area of AI adoption in Aotearoa.

As of 16 September 2026, the J·Index New Zealand dataset recorded 31 verified, source-linked AI implementations across 17 news organisations between 2019 and 2026. The cases include NZME, Stuff, RNZ, TVNZ, Te Hiku Media, Pacific Media Network, The Central App and other publishers.

The most common documented uses are:

Transcription and translation.
Research and document analysis.
Metadata and search optimisation.
Content transformation and summarisation.
Audio and visual production.
Recommendation and curation.
Limited automation of routine or data-driven stories.

The dataset records a relatively high proportion of transcription and translation cases compared with the global sample. That is significant in Aotearoa, where multilingual journalism, te reo Māori and Pacific-language broadcasting create practical use cases that are not reducible to generic content generation. jindex.ai

Recent examples include:

Pacific Media Network: a custom Gemini workflow translates daily news bulletins into Tongan and Cook Islands Māori. The reported time saving remains vendor- and participant-reported, and the planned extension to additional regional languages is not yet evidence of completed scale-up.
The Central App: an AI workflow identifies possible stories in local-council meeting transcripts. Google reports that two reporters doubled their output, but the result has not been independently audited.
Te Waha Nui: AI was used for interview transcription, research support and parts of a te reo Māori translation workflow, with staff checking the output and disclosing the use to readers.
Te Hiku Media: the Kaituhi system provides automatic speech recognition and editable transcription for te reo Māori material, with human correction before publication or captioning.

The pattern is consistent: AI is most defensible where it handles a bounded task and a journalist, broadcaster or language specialist remains responsible for accuracy, context and publication.

The evidence does not support a conclusion that New Zealand newsrooms are moving rapidly toward autonomous journalism. It supports a more limited conclusion: AI-assisted newsroom infrastructure is becoming operational in selected organisations.

Marketing and advertising: workflow integration remains ahead of measurement

Marketing continues to show some of the most mature organisational integration. The previous update’s Trade Me example remains relevant: AI was reported across briefing, strategy, creative development, modelling, optimisation, reporting and anomaly detection.

The important feature is not the number of tools but the breadth of workflow coverage. Marketing teams are using AI across planning and performance functions, while human staff retain responsibility for brand positioning, cultural fit, approvals and risk.

Public evidence remains self-reported. There is still no independent New Zealand-wide measurement of whether AI has reduced costs, increased campaign effectiveness, changed agency staffing or improved creative quality. The sector has moved further toward routine use, but not toward transparent benchmarking.

The same distinction applies to AI-mediated discovery. New Zealand creative organisations increasingly need to consider how brands, stories and cultural content are summarised and recommended by search engines, chatbots and agents. That creates new work around structured information, metadata, authority and provenance, but public evidence of widespread agentic marketing deployment remains thin.

Screen, film and VFX: formal governance is ahead of production evidence

The screen sector continues to have the clearest public governance framework.

NZ On Air’s AI Content Creator Guidance requires applicants proposing AI use to explain how it will affect development, production or distribution. Assessment includes cultural authenticity, creative integrity, consultation, rights, technical capability and the role of human judgement.

The guidance explicitly states that AI should support rather than replace human creativity. Applicants remain responsible for permissions, intellectual property and legal compliance; NZ On Air is not acting as an industry-wide rights regulator. nzonair.govt.nz

NZFC’s AI guidance and funding requirements follow a similar direction, particularly around te ao Māori, mātauranga Māori, disclosure and human creative leadership. The latest public-sector development is therefore governance by funding design, not regulation of every production tool.

Public evidence of operating AI use in commercial New Zealand productions remains limited. The Aotearoa 1 Minute AI Film Festival, the AI and Creativity Summit and September’s WIFT NZ discussion show active experimentation and professional debate. They do not demonstrate that generative production has become a normal production model across the screen sector.

The 30 September decision by NZFC to end its Core Funding Programme from 30 June 2027 is not an AI policy, but it will change the institutional environment in which screen organisations develop capability, advocacy and professional standards. NZFC says targeted industry-development funding will continue. nzfilm.co.nz

Music: assistive use is plausible; commercial rights remain unsettled

Music creators are likely to continue using AI for low-risk or assistive tasks such as:

Idea development and arrangement.
Editing and production assistance.
Translation and accessibility.
Promotion and audience analysis.
Administration and rights management.

The boundary is more contested where AI generates voices, imitates artists, produces substantial musical material or uses copyrighted recordings for training.

NZ On Air’s current music-funding process asks applicants to explain proposed AI use and directs them to its AI guidance. The guidance assesses cultural authenticity, creative integrity, distinctiveness, transparency and technical capability, but does not determine copyright ownership or enforce rights management. nzonair.govt.nz

The September Recorded Music NZ position paper strengthens the rights-holder position described in the previous update. It seeks licensing and training-data transparency rather than a blanket ban on artists using assistive tools. That distinction is likely to remain central to future policy.

There is still no public New Zealand dataset showing how many released songs contain AI-generated elements, how often synthetic voices are used, or whether AI has changed the economics of local music production.

Games: strong technical incentives, little AI-specific reporting

The games sector has the strongest structural incentives to use AI in areas such as prototyping, localisation, quality assurance, asset production, animation, simulation and production management.

The September Games Quarterly reports that the 2026 Game Development Sector Rebate supported 43 studios, with combined reported revenue of $829 million and 194 games in development. Ninety percent of recipient studios are small or medium-sized. These figures demonstrate a substantial digital-production base, but the rebate data does not identify which studios use AI or how AI affects employment, revenue, output or intellectual property. nzonair.govt.nz

The evidence gap is particularly important in games because AI use may be embedded inside commercial development software and therefore invisible in public reporting. The sector may be adopting AI more deeply than the available evidence suggests, but that remains an inference rather than a measured finding.

Independent arts, design and cultural practice: visible experimentation, limited scale

Independent arts and design have the least systematic public evidence of adoption.

September’s Algorithm & Aesthetic II exhibition at Whitecliffe brought together artists, designers and researchers examining AI, authorship, originality, automation and human intervention. Its use of risograph printing deliberately contrasted digital generation with physical process and material imperfection.

This is useful evidence of creative inquiry and public engagement, not evidence of commercial deployment. Similar caution applies to AI film festivals, workshops and artist talks: they show experimentation, skills development and debate, but not necessarily durable business models or scaled production. whitecliffe.ac.nz

Workforce effects: the strongest concern is at the entry level

The most specific recent workforce evidence comes from a Massey University pilot study. The study involved ten one-hour interviews with people across filmmaking, music, design agencies and digital art.

Its findings were mixed:

Interviewees were concerned about displacement but could not identify specific examples of paid creative roles already being eliminated by generative AI.
AI was being used for low-value administrative work.
Some low-paid, freelance and unpaid tasks that help people enter the industry were being replaced.
Participants reported time savings but generally no noticeable impact on profitability.
Most would not deploy AI outputs commercially without extensive human review.

Because the sample is small and qualitative, it cannot establish sector-wide employment effects. It does, however, identify a plausible mechanism of harm: AI may remove the early-career work through which emerging creatives build portfolios, relationships and judgement before it replaces established creative occupations. massey.ac.nz

Governance, Policy and Regulation

Copyright remains unresolved

New Zealand’s current copyright framework has not settled the central generative-AI questions.

The Government’s Copyright Act update confirms that Cabinet has asked the Minister of Commerce and Consumer Affairs to report by 31 March 2027 on a possible copyright framework for generative AI.

The unresolved issues include:

Whether AI developers may train on copyright works without permission.
Whether creators can opt out or require licensing.
Whether training-data records must be disclosed.
Whether AI-assisted and AI-generated outputs qualify for copyright.
Who owns works created with substantial AI involvement.
How voice, likeness and style imitation should be treated.
How Māori cultural knowledge and collective interests should be protected.

The June 2026 copyright package strengthened several existing creator and cultural-institution settings, but it did not resolve generative-AI training or output rights. The Recorded Music NZ and PANZ positions therefore seek to influence a policy process that remains open. mbie.govt.nz

Public funding remains the most immediate governance mechanism

For many practitioners, the most immediate AI requirements are set by funders rather than legislation.

NZ On Air and NZFC ask applicants to explain:

The purpose and extent of AI use.
The human creative contribution.
Rights, permissions and data sources.
Cultural consultation.
Disclosure and audience transparency.
Technical and quality controls.
Potential effects on the creative workforce.

This is a practical model for a small creative economy. It allows funding agencies to require accountability without attempting to regulate every commercial tool.

Its limitation is coverage. Public funding conditions do not apply to every private commission, advertising campaign, independent release or software workflow. They also place documentation demands on small organisations that may have limited legal or administrative capacity.

Māori and Pacific cultural governance is becoming a practical adoption condition

The sector’s governance debate is not limited to generic concerns about bias or copyright.

The relevant questions include:

Whether te reo Māori and Pacific languages are represented accurately.
Whether cultural knowledge is being used with permission and appropriate authority.
Whether collective interests can be represented through individual contracts.
Whether AI outputs reproduce stereotypes or decontextualised cultural material.
Whether language and cultural data remain under local control.
Whether communities benefit from systems built using their knowledge and creative work.

PANZ’s position statement, Huanui’s stated research principles and NZ On Air’s funding guidance all point toward a model in which cultural legitimacy is part of technical and professional competence. This is not yet a complete system of Māori data sovereignty or cultural intellectual-property protection, but it is a clear direction of travel. publishers.org.nz

Election policy may accelerate the debate, but not the legal outcome

Labour’s September AI Action Plan proposes an Office of AI, support for small businesses and community organisations, an AI-in-education team, productivity-sharing measures and a copyright framework for creators.

The proposal is politically relevant because the 2026 general election is scheduled for 7 November 2026. It should not be confused with enacted policy, and other parties’ final positions may differ. The immediate practical effect is to make creator compensation, copyright and accountability part of the election debate rather than solely an official review process. labour.org.nz

Case Studies

Pacific Media Network: multilingual translation

Pacific Media Network’s AI translation workflow remains one of the clearest examples of a high-value use case in Aotearoa.

The system translates daily news bulletins into Tongan and Cook Islands Māori. Google reports that manual translation time fell from as much as three hours to approximately 15 minutes. Human journalists and broadcasters remain responsible for checking the material.

The case demonstrates why language and accessibility applications may be among the most valuable uses of AI for New Zealand creative organisations. It also shows the limits of current evidence: the time-saving figure is self-reported, and public information does not yet establish the quality, cost or audience effects of regional expansion. jindex.ai

The Central App: extracting stories from public documents

The Central App uses AI to identify potential story angles in local-council meeting transcripts. Reporters use the suggestions to find and pursue stories before official livestreams are released.

This is a strong example of AI addressing an information bottleneck rather than replacing editorial judgement. The value comes from finding relevant material in a large document set; the reporter still decides whether the issue is newsworthy, verifies the facts and develops the story.

Google’s reported doubling of output is not independently audited and does not show whether the additional stories generated greater audience reach, revenue or public value. jindex.ai

Te Waha Nui: disclosed assistive use

Te Waha Nui used AI for interview transcription, research support and parts of a te reo Māori translation workflow. Staff checked the output and disclosed the use to readers.

The case is important because it combines three features that are likely to become standard:

A bounded assistive task.
Human checking by people with relevant editorial and language expertise.
Disclosure where AI materially contributed to the work.

This is a more defensible model than presenting AI-generated material as wholly human-produced, particularly for stories involving language revitalisation and cultural knowledge. jindex.ai

Trade Me: integrated marketing use

Trade Me’s Marketing Guild remains the clearest publicly described example of AI being integrated across a commercial marketing workflow.

AI is reported to support briefing, strategy, modelling, creative development, optimisation, reporting and anomaly detection. The organisation’s “co-pilot, not autopilot” framing captures the current mainstream adoption model: AI increases speed and analytical capacity, while people retain responsibility for judgement, brand and risk.

The account is self-reported. It does not establish audited cost savings, changes in headcount or improvements in campaign performance.

Whitecliffe: creative practice as critique

Whitecliffe’s Algorithm & Aesthetic II exhibition provides a useful counterpoint to operational case studies. Rather than presenting AI simply as a productivity tool, the participating artists and researchers examined authorship, originality, automation and human intervention.

The exhibition shows that AI is entering creative education and public discourse as a subject of critique as well as a production technology. It should not be counted as evidence of scaled industry adoption, but it indicates that cultural responses are developing alongside technical experimentation. whitecliffe.ac.nz

Trends

1. Operational AI remains ahead of synthetic public-facing content

The strongest evidence continues to concern translation, transcription, research, summarisation, metadata, accessibility, administration and optimisation.

These tasks are bounded, relatively easy to review and often address clear labour bottlenecks. Fully synthetic journalism, music, film, literature or visual art remains more difficult to defend where provenance, originality, cultural meaning and audience trust are central.

2. The sector is moving from tool experimentation to workflow governance

The maturity question is increasingly not “does this organisation use AI?” but:

Which tools are approved?
What data can be supplied?
Who checks the output?
When is use disclosed?
How are prompts, source materials and edits recorded?
How are rights and cultural permissions established?
Who benefits from productivity gains?

This shift is visible in funder questionnaires, newsroom principles, publisher statements and sector position papers. Formal governance remains uneven, especially among sole traders and small studios, but it is becoming part of ordinary professional practice.

3. Rights policy is becoming part of competitiveness

The music and publishing positions show that rights are not only a compliance issue. They may become a competitive differentiator.

Creators and organisations with clear provenance, consent and licensing processes may be better positioned with funders, commissioners, platforms, audiences and international partners. Conversely, systems that cannot explain where training material came from may face reputational, legal and commissioning risk.

4. Aotearoa’s distinctive opportunity is language and cultural capability

Translation, transcription, captioning, language learning, archival discovery and accessibility are areas where New Zealand organisations can develop locally relevant systems.

The opportunity is not simply to apply global models to local content. It is to build processes that respect language variation, cultural authority, community consent and Māori and Pacific data interests. That requires more than technical accuracy. It requires governance by the people whose knowledge and identity the systems represent.

5. Entry-level creative work is a more immediate concern than mass professional displacement

The Massey pilot does not show widespread job losses. It does show that AI can remove low-paid or unpaid tasks before replacing established occupations.

That distinction matters for policy and education. If junior researchers, assistants, production runners, copy editors, designers and freelance contributors lose access to early work, the industry may experience a delayed capability shortage even if current productivity improves.

6. Capability programmes are multiplying faster than evaluation

The Government’s small-business pilot, Applied Intelligent Systems subject, Huanui, AI Forum workstreams and proposed national creative AI research platform all indicate expanding capability investment.

What remains missing is systematic evaluation:

Which programmes lead to sustained use?
Which tools produce measurable value?
Are benefits retained by creative workers or captured by platforms and clients?
Do Māori and Pacific practitioners have equal access?
Does AI strengthen local creative ownership or increase dependence on offshore providers?

Until these questions are measured, the sector should distinguish carefully between capability-building and adoption.

7. Evidence remains biased toward visible organisations

Large media companies, technology partners, universities, public funders and industry bodies are more likely to publish AI case studies. Independent artists, small studios, Māori-led organisations and Pacific practitioners often have less capacity or incentive to document their use.

The lack of public evidence outside journalism and marketing therefore does not prove that adoption is absent. It means that claims about scale, productivity and workforce effects should remain cautious.

Outlook

The next six to twelve months are likely to be shaped by:

The 7 November 2026 general election and competing proposals on creator rights, AI governance and small-business support.
Continued public debate over the Recorded Music NZ position paper and related copyright demands.
Further guidance or policy development before the Government’s 31 March 2027 deadline for advice on a generative-AI copyright framework.
Publication of results from the 2026 Survey of Business Operations.
Possible clarification of whether the Aotearoa Creative AI Research Institute becomes a funded and operating platform.
More disclosure and provenance requirements in public funding and commissioning.
Further testing of AI translation and transcription in Māori and Pacific-language media.
Continued pressure to measure effects on freelance and entry-level creative work.
Greater use of AI for audience discovery, metadata, localisation and accessibility.
More negotiations over synthetic voices, digital replicas, likeness, style imitation and training-data records.

The practical question for creative organisations is becoming less whether to use AI and more where it creates defensible value. The strongest candidates are workflows where the problem is repetitive, the output can be checked and the use does not compromise authorship, cultural authority or audience trust.

Overall Assessment

AI adoption in Aotearoa New Zealand’s creative industries has progressed, but mainly through documented workflow use and stronger governance rather than scaled synthetic production.

Journalism and publishing remain the best-evidenced subsectors. Marketing shows broad organisational integration, although its results are mostly self-reported. Music has developed the clearest rights-based policy response. Screen and public funding have the most mature formal governance. Games and VFX have strong technical incentives but limited AI-specific data. Independent arts and design show active experimentation but little evidence of durable commercial deployment.

The latest evidence also makes the labour issue more precise. The immediate risk is not necessarily the rapid disappearance of established creative professions. It is the erosion of the small, low-paid and informal tasks through which new practitioners enter the industry.

New Zealand’s likely advantage lies in practical, culturally grounded applications: multilingual media, accessibility, archival discovery, local storytelling, creative technology and rights-aware production systems. Capturing that advantage will require more than access to global AI tools. It will require evidence, consent, provenance, cultural governance, workforce investment and a fair distribution of productivity gains.

The central assessment remains unchanged: the Aotearoa creative sector is adopting AI, but it is doing so cautiously, unevenly and under increasing pressure to prove that the technology strengthens rather than extracts from the creative ecosystem.