AI in Creative Industries in Aotearoa New Zealand: A Living Whitepaper

AI use across Aotearoa New Zealand’s creative industries is becoming routine in workflows and distribution, but evidence of scaled generative production remains limited. Governance, consent, cultural integrity and entry-level work now shape adoption as much as efficiency.

Executive Summary

The latest evidence points to broad but uneven AI adoption. Generative AI is already common among digitally active creators, but use is concentrated in ideation, transcription, translation, summarisation, optimisation, accessibility and administration rather than fully automated public-facing production.

The strongest evidence of operating deployment is in journalism and publishing. Google-supported pilots report faster multilingual translation, increased story production and improved search capability, although the results are primarily self-reported and have not been independently audited. Marketing teams are also embedding AI across planning, creative development, modelling, optimisation and reporting.

By contrast, evidence remains thin for independent arts, music production, screen production and games. These sectors are clearly experimenting and developing governance responses, but public reporting rarely distinguishes between a funding announcement, a limited pilot and a system operating at scale.

Several developments since the previous update are significant:

• The music industry has made consent, licensing and training-data records part of its 2026 election-policy platform.
• AI discovery and agent-mediated search are becoming commercial strategy issues, not simply content-production issues.
• Public funding agencies are turning AI disclosure, cultural authenticity and human creative leadership into routine assessment requirements.
• AI capability is entering the vocational education pipeline through a new industry-led Applied Intelligent Systems subject.
• The national evidence base remains incomplete. The 2026 Survey of Business Operations includes AI adoption and impact questions, but results are not yet available.

The emerging Aotearoa model is therefore human-led, workflow-oriented and increasingly governed upstream through funding, commissioning, education and rights policy.

What Has Changed Since the Last Update

Music rights have moved from sector concern to election-policy demand

On 25 August 2026, 12 organisations from the Aotearoa music community published a 2026 Music Manifesto. Its AI and copyright demands are among the clearest sector positions yet published in New Zealand.

The manifesto calls on the next Government to:

• Introduce no new copyright exceptions allowing AI companies to train on music without permission.
• Preserve artists’ and rights holders’ ability to choose how their work is used.
• Require AI companies to keep meaningful records of the music used for training.
• Provide those records to rights holders on request.
• Treat music as a growth and export industry alongside screen and games.

The document does not oppose musicians using AI as an assistive creative tool. Its distinction is between creator-controlled use and commercial systems trained on copyrighted music without consent. This is a more precise position than a general pro- or anti-AI stance, and it is likely to shape political debate during the 2026 election cycle. (apraamcos.co.nz)

Newsroom pilots are beginning to acquire operating pathways

The Google News AI Workshops remain the clearest recent examples of AI moving beyond experimentation into repeatable newsroom workflows.

Google reports that:

• Pacific Media Network reduced manual translation of daily bulletins into Tongan and Cook Islands Māori from up to three hours to approximately 15 minutes. Google says the tool is being rolled out regionally in September 2026.
• The Central App used an AI tool to identify story angles in council meeting transcripts. Two reporters reportedly doubled their story output.
• Newsroom NZ used an AI tool to help reporters produce search metadata, with self-rated SEO confidence rising from 2.7 to 4.3 out of five.

These are meaningful operating signals, but they remain vendor- and participant-reported results. The evidence establishes that the tools are being used in live workflows or pilots; it does not establish sector-wide productivity gains, financial returns or long-term editorial effects. The wider 18-month Google programme for New Zealand and Pacific publishers is an announced capability programme, not evidence that all participating organisations are operating AI systems at scale. (blog.google)

AI discovery is becoming a commercial creative issue

The IAB New Zealand Discovery: AI & Search Summit, held in August and summarised on 19 August, framed AI-generated answers, agent-mediated search and changing consumer behaviour as central issues for advertisers, agencies, publishers and media owners.

The significance is strategic. Creative organisations increasingly need to consider not only how content is produced, but how it is represented, retrieved and recommended by AI systems. This creates a new layer of work around structured content, brand authority, metadata, search visibility and the accuracy of AI-generated summaries.

The summit itself is evidence of sector attention and capability-building, not evidence that New Zealand businesses have broadly implemented agentic marketing systems. (iab.org.nz)

The adoption baseline has become more useful, but not more granular

The latest national creator data remains Manatū Taonga’s 2025 Cultural Participation Survey. It found that 69% of creators use digital tools, representing 28% of adults, and that 65% of those digitally active creators use generative AI.

The denominator matters. The figure does not mean that 65% of all New Zealand creators or adults use generative AI. It indicates substantial adoption among a digitally active subset. The main uses remain exploratory and assistive:

• 49% use generative AI to explore or improve ideas.
• 34% use it to generate or produce creative work.
• 14% use it to share work more widely or improve accessibility.

Among creators who do not use digital tools, 36% identify a lack of skills or knowledge as a barrier. (mch.govt.nz)

There is still no new public evidence of scaled AI production in several subsectors

Since the previous update, publicly available material has not established a major operating deployment in independent visual arts, music production, screen production or games comparable to the newsroom examples.

The developments in these areas remain primarily:

• Governance requirements.
• Training and capability programmes.
• Creative experiments and competitions.
• Proposals for research infrastructure.
• Advocacy over copyright, consent and cultural rights.

That does not mean adoption is absent. It means that much of it is private, embedded inside commercial software, or not publicly documented.

Current State of AI Adoption

National creative practice: broad experimentation, limited measurement

Aotearoa has a relatively strong measure of individual creator use but a weak measure of organisational adoption. The Cultural Participation Survey captures whether creators use digital tools and generative AI, but not the maturity of their workflows, the scale of commercial deployment, the effect on employment or whether AI-generated outputs reach audiences.

The 2026 Survey of Business Operations, being conducted by MBIE and Stats NZ, includes questions on AI adoption and impact. Its results should improve the national evidence base, but no findings are yet available. Until then, claims about AI adoption in creative businesses should be treated as indicative rather than comprehensive. (mbie.govt.nz)

Journalism and publishing: the most mature documented use

Journalism remains the best-documented New Zealand creative subsector for operational AI.

The AUT Journalism, Media and Democracy baseline report identifies AI use in:

• Searching and research.
• Transcription.
• Summarising long documents.
• Spelling and grammar checking.
• Content transformation.
• Audio and video generation.
• Homepage curation and recommendations.
• Drafting or assisting with some news articles.

Publicly documented examples include NZME’s BusinessDesk use of AI to generate articles from NZX announcements, Stuff’s use of AI for first drafts from single-source material, and AI-assisted curation and recommendation within digital news products.

The dominant pattern is not autonomous journalism. It is structured automation around repeatable, low- to medium-complexity tasks, followed by human checking or editorial control. (aut.ac.nz)

The main constraint is trust. AUT’s 2026 Trust in News research found that 60% of New Zealanders remain uncomfortable with news mostly produced by AI, even with some human oversight. RNZ’s principles generally prohibit knowingly publishing or broadcasting generative-AI-created material, while permitting assistive use for research, brainstorming, administration, summarisation and transformation of existing content. TVNZ similarly requires human accountability, risk assessment and disclosure where AI materially affects editorial or audience-facing content. (aut.ac.nz)

Advertising and marketing: institutionalising faster than other creative fields

Commercial marketing appears to be the fastest-normalising area of AI adoption.

The Marketing Association’s Digital Day Out report describes Trade Me’s Marketing Guild as using AI across:

• Briefing and strategy.
• Creative development.
• Data modelling.
• Copy optimisation.
• Performance reporting.
• Anomaly detection.

Trade Me’s stated principle was “co-pilot, not autopilot”. This is consistent with the broader evidence: AI is being positioned as an operating layer across the marketing lifecycle rather than simply as a copywriting tool.

The evidence is still largely self-reported. The Trade Me account demonstrates organisational integration, but it does not provide an independently measured comparison of productivity, campaign effectiveness, staffing or creative quality. Similarly, a trade-press report on Xero’s AI-powered global campaign describes the creation of 600 campaign variations across eight markets in eight weeks. That is evidence of significant production scale, but it is not a New Zealand-only deployment and the reported results come from the company and agency involved.

The sector’s next challenge is discoverability. AI systems may become an important route through which audiences encounter brands, products, publishers and cultural content. This makes factual accuracy, structured information, brand distinctiveness and rights-cleared source material part of creative competitiveness.

Screen, film and VFX: governance is more advanced than deployment evidence

The screen sector has the most developed formal AI governance among publicly funded creative fields.

The New Zealand Film Commission’s June 2026 funding guidance requires applicants to outline proposed AI use. Applications are assessed for:

• Cultural authenticity and appropriate consultation.
• Respect for te ao Māori and mātauranga Māori.
• Preservation of human creative leadership.
• Transparency about AI’s role.
• Legal and ethical implications.
• Wider effects on the screen workforce and industry.

NZ On Air’s AI Content Creator Guidance takes a similar approach. It asks applicants to explain how AI supports rather than replaces human creativity, identify risks, demonstrate technical capability and disclose the role of AI in content creation. NZ On Air explicitly notes that it does not enforce copyright law; responsibility remains with applicants, producers and commissioners. (nzfilm.co.nz)

This is a significant shift in practice. AI is no longer only a production question. It is now part of funding assessment, cultural risk management and accountability to audiences.

However, governance should not be confused with adoption. The Aotearoa AI and Creativity Summit, FutureFrames and the One Minute AI Film Festival demonstrate active experimentation and skills development. Public evidence of AI being used across large numbers of commercial New Zealand productions remains limited.

Music: assistive use is accepted, but commercial rights remain contested

Music has a mixed adoption profile. The APRA AMCOS research on AI and music indicates that many Australasian music creators are early users of AI tools and that more than half believe AI can assist human creativity. At the same time, 82% expressed concern about their ability to earn a living, and the research projected that 23% of music creators’ revenues could be at risk by 2028. These are Australasian survey findings and projections, not a current New Zealand-only measurement.

New Zealand’s immediate response has been governance rather than prohibition. NZ On Air’s music-funding process asks applicants to explain how AI or generative AI will be used in song creation. Its criteria include cultural authenticity, creative integrity, transparency, technical capability, accessibility and industry impact.

The 2026 Music Manifesto adds a stronger rights position: no unlicensed AI training on music, meaningful records of training material and a right for creators to choose whether their work is used. The practical adoption model is therefore likely to remain:

• Private experimentation.
• AI-assisted songwriting and production.
• Tools for promotion, translation, accessibility and audience analysis.
• Human-led releases where rights and provenance can be demonstrated.
• Caution around synthetic voices, imitation and fully generated recordings.

Games and creative technology: strong economic incentives, weak AI-specific data

New Zealand’s games sector continues to provide a strong environment for AI experimentation. NZ On Air’s 2026 Game Development Sector Rebate data records 43 recipient studios, $829 million in combined revenue, 194 games in development and 98.3% export revenue. Ninety percent of recipient studios are SMEs. (nzonair.govt.nz)

These conditions create incentives to use AI for:

• Prototyping.
• Asset generation.
• Localisation.
• Testing and quality assurance.
• Animation and rigging.
• Simulation.
• Production management.

But the GDSR data does not identify which studios use AI or measure its contribution to revenue, staffing or output. It is therefore evidence of sector scale and technical capacity, not evidence of AI adoption itself.

Games and VFX may ultimately generate some of the largest productivity gains because their workflows contain substantial volumes of repeatable digital production. They also face some of the most difficult questions about training data, originality, artistic labour and the preservation of distinctive human-made worlds.

Arts, cultural institutions and education: capability is becoming a policy priority

Independent arts and cultural institutions have less publicly documented AI deployment than media and marketing. The main opportunities identified in government research are accessibility, transcription, translation, collection discovery, digital preservation and assistance with administration.

Manatū Taonga’s Culture in the Digital Age Long-term Insights Briefing places AI within a longer-term cultural system shaped by Māori data sovereignty, te reo Māori, cultural stewardship, sustainability and unequal access to technology. It also warns that accessibility tools need testing with affected communities and should not be treated as substitutes for accessibility by design. (mch.govt.nz)

Workforce development is beginning to reflect this shift. On 6 August 2026, the Electrotechnology, Information Technology, and Creative Industry Skills Board announced that it would lead development of Applied Intelligent Systems for the senior secondary system. The subject is intended to cover machine learning, agentic systems, human oversight, ethical reasoning and the design of intelligent workflows across industries including creative practice.

This is an important capability signal, but it remains an announcement and development process rather than evidence that the new subject is already being widely taught.

Governance, Policy and Regulation

Copyright remains unresolved

New Zealand still does not have a dedicated generative-AI copyright framework.

MBIE’s Copyright Act update states that Cabinet has invited the Minister of Commerce and Consumer Affairs to report by 31 March 2027 on a possible framework for generative AI.

The Government’s current copyright package addresses Free Trade Agreement obligations and other targeted reforms, but it does not settle:

• Whether AI developers may train on copyright works.
• Whether creators can opt out or license training use.
• Whether AI-generated outputs qualify for copyright.
• Who owns AI-assisted works.
• How voice, likeness and style imitation should be treated.
• How collective cultural knowledge should be protected.

The music industry’s election manifesto therefore arrives while the law remains open. Its demand for no new AI-training exceptions is a policy position, not current law. (mbie.govt.nz)

Public funding is becoming a practical governance lever

The most immediate AI regulation affecting creative practitioners is not legislation. It is the way public funders assess applications.

NZFC, NZ On Air and Creative New Zealand all expect applicants to disclose AI use and take responsibility for:

• Human authorship and creative direction.
• Cultural integrity.
• Privacy and intellectual property.
• Appropriate consultation.
• Audience transparency.
• Risk management.
• The effect of AI on the wider creative workforce.

This approach has two effects. It encourages responsible experimentation, but it also creates additional compliance and documentation requirements for small organisations. The likely direction is toward lightweight provenance records: what tools were used, what material was supplied, what decisions remained human and how final outputs were reviewed.

Māori rights and cultural governance cannot be treated as generic ethics

Government guidance increasingly recognises that AI risks in Aotearoa are not limited to individual copyright ownership. Māori communities have raised concerns about the use of collective knowledge, whakapapa, traditional stories, te reo Māori and other taonga in systems controlled by multinational companies.

MBIE’s responsible-AI guidance warns that generative outputs may resemble existing copyright works, recommends licensing assessments and record-keeping, and advises organisations to work with Māori where Māori data or cultural knowledge is involved. The National Intellectual Property Management Policy, updated on 18 August 2026, also requires Treaty, Māori, mātauranga Māori and Indigenous rights and interests to be properly considered in publicly funded research commercialisation. (mbie.govt.nz)

These policies do not create a complete framework for cultural data sovereignty, but they establish a direction: cultural legitimacy is part of responsible AI capability, not an optional communications exercise.

Trust, sustainability and human accountability are converging

AI governance is expanding beyond accuracy and copyright. Public concerns also include job displacement, reduced human interaction, privacy, over-reliance on automated systems and the environmental cost of AI infrastructure.

For creative organisations, this means that a defensible AI deployment increasingly needs to answer four questions:

1. What value does the system add?
2. What human responsibility remains?
3. Whose material, culture or identity is being used?
4. Who benefits and who bears the cost?

Case Studies

Pacific Media Network: translation as an assistive deployment

Google reports that Pacific Media Network built a Gemini-based tool to translate daily news bulletins into Tongan and Cook Islands Māori. The reported reduction from up to three hours to approximately 15 minutes illustrates a high-value use case: AI increases language coverage while human journalists and broadcasters retain editorial responsibility.

The planned regional rollout in September 2026 will be a useful test of whether a successful pilot can operate reliably across different content, dialect and quality-control conditions. Until deployment results are published independently, the reported time saving should be treated as provisional. (blog.google)

The Central App: document analysis for local reporting

The Central App’s AI tool identifies potential story angles in dense council meeting transcripts. Google says two reporters doubled their output and were able to identify stories before official livestreams.

This is a strong example of AI supporting journalism where the bottleneck is not writing but finding relevant information in large public documents. It also illustrates why domain-specific workflows may be more valuable than general-purpose content generation. The claim remains participant-reported and does not show whether the additional stories produced greater audience reach, revenue or public value. (blog.google)

Trade Me: AI embedded across the marketing lifecycle

Trade Me’s Marketing Guild provides a more organisationally mature example. AI is described as being used from strategy and briefing through to modelling, creative development, reporting and anomaly detection.

The case demonstrates that adoption maturity is better measured by workflow integration than by the number of individual AI tools in use. It also supports the view that human judgement remains central in brand positioning, cultural fit, risk and final approval. The public account does not, however, provide audited productivity or campaign-performance data. (marketing.org.nz)

NZ On Air and NZFC: governance by funding design

The most scalable public-sector intervention is the inclusion of AI questions in funding applications.

Rather than banning AI, NZ On Air and NZFC require applicants to explain its purpose, risks, cultural implications, technical controls and effect on human creativity. This creates a governance model suited to a small creative economy: it can be applied before public money is committed, without requiring a separate regulator for every tool or production workflow.

Its limitation is enforcement. Funding agencies can assess applications, but they cannot resolve every underlying copyright, licensing or labour issue.

Trends

1. Workflow AI is ahead of synthetic public-facing content

The strongest evidence concerns translation, transcription, document analysis, metadata, summarisation, research, administration and optimisation. These tasks have clear boundaries and can be checked by human specialists.

Fully synthetic film, music, journalism or visual art remains less defensible where provenance, originality, cultural meaning or audience trust are central.

2. Adoption is shifting from experimentation to operating models

The important change is not simply that more creators have access to AI. It is that organisations are defining:

• Approved tools.
• Human review points.
• Disclosure requirements.
• Data-handling rules.
• Procurement criteria.
• Cultural consultation processes.
• Records of AI involvement.

This is the beginning of organisational maturity, although many small creative businesses lack the time and expertise to formalise these practices.

3. Distribution and discoverability are becoming part of production

AI is changing how audiences find content. Search summaries, recommendation systems and agent-mediated discovery can influence which creators, brands and stories are visible.

For New Zealand content, this raises a particular risk: local work may be produced successfully but remain difficult for audiences to find within global systems. NZ On Air’s 2026 audience research found that 81% of New Zealanders like seeing New Zealand faces and places on television, while 41% would listen to more New Zealand music if it appeared on their streaming service. The problem is therefore not only production capacity; it is discoverability and distribution. (nzonair.govt.nz)

4. Rights and consent are becoming competitive conditions

The 2026 Music Manifesto shows that rights holders are not rejecting technological change. They are seeking control over training inputs, licensing, attribution and remuneration.

Similar concerns apply across screen, advertising, publishing and visual arts, particularly where AI can reproduce a person’s voice, face, style or cultural identity. Organisations with clear provenance and consent processes will be better positioned with funders, commissioners, audiences and international partners.

5. Entry-level work remains the most vulnerable part of the workforce

The immediate labour risk is not necessarily the disappearance of established creative professions. It is the reduction of low-paid, freelance, assistant and unpaid work through which emerging practitioners acquire portfolios, networks and practical judgement.

This risk is difficult to measure because the work is often informal and distributed across small businesses. It also creates a policy tension: AI may lower barriers to entry for some creators while removing the first professional steps for others.

6. The evidence is still biased toward visible, well-resourced organisations

Large media companies, technology partners, funders and industry bodies are more likely to publish AI case studies. Independent artists, Māori-led organisations, Pacific practitioners and small studios are less likely to have the resources to document adoption publicly.

The absence of evidence should not be interpreted as absence of use. It does mean that sector-wide claims should be made cautiously.

Outlook

Over the next 12 months, the most consequential developments are likely to be:

• The September 2026 regional rollout of Pacific Media Network’s translation workflow, which will test whether a newsroom pilot can scale across Pacific-language content.
• Further integration of AI disclosure into NZFC, NZ On Air and Creative New Zealand funding and commissioning processes.
• Publication of MBIE and Stats NZ findings from the 2026 Survey of Business Operations.
• Political debate over the music industry’s demands for consent, licensing and AI training records.
• Continued growth in AI-driven search, recommendation and agent-mediated discovery.
• More use of AI for accessibility, multilingual content and audience segmentation.
• Greater demand for provenance records covering prompts, source material, human edits and final approvals.
• More pressure from creators over digital replicas, voice imitation, synthetic performance and training data.
• Further scrutiny of entry-level employment and whether productivity gains are reaching creative workers.
• The 31 March 2027 deadline for Government advice on a possible New Zealand generative-AI copyright framework.

The main uncertainty is not whether AI capability will improve. It is whether New Zealand’s creative institutions can convert that capability into sustainable creative businesses while retaining local ownership, cultural legitimacy and professional pathways.

Overall Assessment

AI adoption in Aotearoa New Zealand’s creative industries is broad, uneven and increasingly institutionalised.

Journalism and marketing show the clearest evidence of operating use. Screen and public funding have the most developed governance. Music has the strongest organised resistance to unlicensed training and the clearest demand for creator control. Games and VFX have substantial technical incentives but little public AI-specific measurement. Independent arts and cultural organisations remain the least visible in the evidence base.

The sector is not moving uniformly toward autonomous content generation. The dominant model remains human-led, assistive and workflow-oriented.

New Zealand’s distinctive challenge is to ensure that AI adoption strengthens, rather than extracts from, its creative ecosystem. That requires more than access to tools. It requires rights and consent, cultural governance, transparent commissioning, skills development, evidence-based evaluation and a fair distribution of productivity gains.