Te Kete o Karaitiana Taiuru (Blog)

Too often in New Zealand, AI safety folk (who usually have a high level of privilege or are detached from vulnerable communities), will reference international AI stories highlighting harms while being oblivious to local harm stories. This is important from the perspective that New Zealand is not immune from AI harms and it helps politicians to see the issues are not simply isolated New Zealand incidents, but a part of a global issue. Likewise, New Zealand media have statistically referred to only a minority of AI safety experts for media stories, continuing the historical media bias against Māori and Pacific Peoples.

Much like the international media do, New Zealand  media and researchers need to put faces and names to those victims (when they want their voice heard) of AI harms in New Zealand, while calling out government and business who perpetuate these harms in our communities. This both empowers those with no voice and makes it more personal and more easier for people to relate to the harm.

As I did with Te Ani Solomon, the nameless and faceless story of a Māori woman who was falsely accused in a mistaken identity with Facial Recognition Technology; this article introduces another real face, another real name from New Zealand, a victim and her community of deep fake x rated content (images that have been manipulated with AI to create fake images, usually targeting women in naked and or a sexual nature) often called digital rape.

Marae TV on Sunday June 21, 2026 aired Episode 12 aired a story of a Māori community targeted by deepfake images. I couldn’t find any mainstream media articles, but this is normal when minority communities suffer harm. Often, there is a mistrust of mainstream media, or a shame, culturally unsafe experiences with agencies, or maybe a preference to trust a community worker who is bound by confidentiality. In this particular case, I have had no correspondence with the person, this article is based on the public interview on Marae TV.

Marae TV aired the story of a Māori woman called Ario Nepia who was notified by a family member that someone whom Ario stated she had no relationship with, had gone through her Facebook photos, including wedding photos, and created deep fake images of her, her family and community. This is not a unique story and many of the stories I hear about from people in the community are ordinary everyday New Zealand folk, usually teenage and young woman who have been victimized by teenage boys and men.

While the obvious issue is the way we bring up boys, the easy access to x rated content that is typically misogynist and degrading to women, applying solutions to these issues requires a significant social change and changes to the education system in addition to new legislation to make such offenses illegal to apply pressure to both the perpetrators and big tech platforms,  and to empower victims, their families and communities.

The legislative issue

New Zealand currently has no deepfake specific offense, which means victims and prosecutors are left forcing these harms into laws never built for synthetic media. Harassment claims under the Harmful Digital Communications Act move slowly and require mediation first; intimate visual recording offenses often assume a real recording was made, not a fabricated one; and defamation requires a false statement of fact, which doesn’t capture a purely sexual or degrading fabrication that makes no factual claim at all. Without statute, basic questions such as distribution, whether intent to harm must be proven, whether satire is exempt get argued from first principles each time rather than answered in advance.

The practical cost of that gap is a lack of any clear obligation on platforms to remove content quickly. Victims are stuck negotiating with trust and safety teams under voluntary community standards instead of a statutory take down deadline. That gap also compounds harm unevenly voice-cloning fraud and executive impersonation scams, political deepfakes near elections, and impacts on Pacific peoples, rural and low-income communities, and older New Zealanders all worsen where there’s no clear deterrent or fast remedy.

Proposed Bill

ACT MP Laura McClure has proposed a bill to Parliament “Deepfake Digital Harm and Exploitation Bill 2026 ” to address this issue. Public submissions were open and closed on June 19 at just before midnight.

The bill is best summarised in the Bills General policy statement:

“Worldwide there has been a rise in the proliferation of sexually explicit deepfakes. This is where images or videos are digitally altered or synthesised to take an individual’s recognisable likeness and turn it into sexually explicit material.

This form of material constitutes a clear violation of self-ownership and personal autonomy. Just as the unauthorised sharing of genuine intimate imagery is a rights infringement, so too is the false representation of a person in intimate contexts without consent. This conduct misappropriates a person’s image for exploitative purposes, causing reputational, psychological, and often material harm. Criminalising such acts and enabling victims to seek redress and removal aligns with a commitment to defending individuals from coercion, fraud, and aggression.

This Bill holds accountable those who use digital tools to violate other individuals’ agency. It does this by amending both the Crimes Act 1961 and the Harmful Digital Communications Act 2015 to expand the definition of an “intimate visual recording” to explicitly include images created, synthesised, or altered to show a person’s likeness produced without consent. “

I fully support the Bill and that it is implemented urgently to protect people from this despicable behavior. I also feel this is an opportunity to pass the Bill and then address the wider issues with urgency.

My submission TO THE SOCIAL SERVICES AND COMMUNITY COMMITTEE Deepfake Digital Harm and Exploitation Bill 2026

June 18, 2026

Introduction

I welcome the opportunity to submit on the Deepfake Digital Harm and Exploitation Bill. I support the Bill’s core intent to close a real and documented gap in New Zealand law and I commend ACT MP Laura McClure for progressing this legislation.
I write as a Māori technology ethicist with specialist expertise in AI governance, Māori data sovereignty, and Te Tiriti o Waitangi applied to emerging technologies. My work spans AI governance, digital identity, and the rights of tangata whenua and vulnerable communities in New Zealand.

My submission supports the Bill proceeding while pressing the Committee to address significant structural gaps. The harms deepfake technology enables are not evenly distributed. They fall disproportionately on Māori, Pacific peoples, disabled people, LGBTQ+ communities, older people, people in rural and low-income communities, and those with limited access to legal recourse. The Bill as drafted does not adequately acknowledge this reality. It also fails to address the full spectrum of deepfake harm: the Bill focuses on sexually explicit imagery while leaving financial fraud, political manipulation, and democratic disinformation entirely outside its scope.
A Bill that addresses only the most visible form of harm, and only for those best positioned to access its remedies, is a Bill that works for some New Zealanders but not all.

1. Disproportionate Harm to Vulnerable Communities

Research is unambiguous: deepfake image-based abuse does not harm people equally. Communities experiencing multiple marginalised identities including LGBTQ+ individuals, ethnic minorities, disabled people, and low-income and rural communities who face deepfake harm at elevated rates and encounter additional and compounding barriers to seeking support and legal redress. The Bill as drafted treats all victims as equivalently positioned, but they are not.

2.1 Māori
Māori have consistently been subjected to the misappropriation, sexualisation, and distortion of their likenesses and cultural identities from colonial-era photography that framed Māori bodies as ethnographic objects, to contemporary commercial misuse of moko kauae, moko, models, and cultural imagery in digital contexts. Deepfake technology accelerates, scales the abuse.
The Waitangi Tribunal’s WAI 2522 inquiry has established that Māori data (imagery included) is subject to Māori data sovereignty principles grounded in tino rangatiratanga. A deepfake fabrication of a Māori person’s likeness is not only an assault on that individual; it is also an interference with their whakapapa, their mana, and the interests of their marae and hapū in that person’s identity and representation.

For Māori, harm to an individual’s likeness carries dimensions of collective harm that the Bill does not currently recognise. Whakapapa is a foundational organising principle of Māori identity: a person’s face, voice, and bodily image are not simply personal attributes but manifestations of whakapapa, carriers of mana, and extensions of collective identity.

2.2 Pacific Peoples
Pacific peoples in New Zealand share many of the same vulnerabilities to deepfake harm as Māori: strong relational and community-based identity structures, where harm to an individual’s digital likeness carries reputational and relational consequences for extended family and community networks. Many Pacific communities also have deeply held values around modesty, sexuality, and gender expression grounded in their cultural and faith traditions, making the fabrication of sexually explicit imagery of community members particularly devastating. The combination of those values with cultural barriers to reporting, and underrepresentation in digital literacy resources, means Pacific victims are less likely to access the remedies this Bill creates. Implementation must be designed with Pacific communities, not just for them.

2.3 LGBTQ+ Communities
LGBTQ+ individuals consistently report the highest rates of online harm of any demographic group. Research by the Centre for International Governance Innovation found that LGBTQ+ individuals reported the highest proportion of experienced online harm incidents across 18 countries. Deepfakes targeting LGBTQ+ individuals carry a specific additional dimension: they are frequently weaponised to out people, to fabricate sexual content that is then used to reveal, imply, or force disclosure of a person’s sexual orientation or gender identity without their consent. In contexts where a person has not chosen to disclose their identity to family, employer, community, or faith group, such a deepfake can cause catastrophic and irreversible harm. The risk of forced outing through deepfake content is not addressed anywhere in the Bill’s framework. I recommend the Committee direct that guidance materials and victim support resources specifically acknowledge this dimension.

2.4 Disabled People
Disabled people face elevated risks from deepfake abuse for several reasons. Those with cognitive or communication disabilities may be less able to identify that a deepfake has been created of them, to navigate complaint processes, or to access online removal tools. Those with visible disabilities may be targeted with content that exploits or degrades their physical presentation in intimate imagery. And the digital divide remains significant: disabled people are overrepresented in communities with limited digital access and literacy, meaning that remedies dependent on online engagement are effectively inaccessible. The Bill should be accompanied by accessibility commitments ensuring that complaint, reporting, and content removal pathways are fully accessible to disabled New Zealanders, consistent with the New Zealand Disability Strategy.

2.5 Rural and Low-Income Communities
Access to legal recourse requires resources: legal advice, digital literacy, knowledge of available mechanisms, and the practical capacity to navigate court processes. These are not evenly distributed. Rural communities face compounding disadvantage through limited access to legal services, lower digital connectivity, and greater distances from support organisations. Harmful content can spread rapidly through local networks in small communities, causing acute reputational damage before any remedy can be applied. Low-income New Zealanders face similar barriers to access. A Bill that creates rights only meaningfully accessible to those with legal and digital resources replicates systemic inequity. Implementation must include funded, accessible support pathways that reach communities currently underserved by the justice system.

2.6 Older New Zealanders
While the intimate imagery focus of the Bill is most acute for younger people, older New Zealanders face deepfake harm through a different but equally serious vector: financial exploitation. Voice cloning and video deepfakes are now routinely deployed in scams targeting older people, the so-called ‘grandparent scam’, in which a cloned voice purporting to be a grandchild or family member calls claiming an emergency requiring immediate funds. BNZ has identified voice cloning as one of the top AI-related scam concerns in New Zealand, and sextortion scams which use AI-generated deepfake images as leverage to extract payment continue to rise in New Zealand. Older people with limited familiarity with AI capabilities are particularly vulnerable. The Bill does not address these harms at all.

2. Financial Fraud: The Absent Harm

The Bill is silent on deepfake-enabled financial fraud. In New Zealand, direct losses from scams and fraud now exceed NZ$5 million per quarter, and deepfake technology is a primary driver of the most sophisticated and damaging fraud vectors currently active in this country.
Deepfake financial fraud operates across multiple modes. Voice cloning enables criminals to impersonate trusted individuals, family members, employers, bank staff, or government representatives using only seconds of recorded audio scraped from social media or public sources. A reconstructed voice with the person’s cadence, accent, and speech patterns can direct family members to transfer emergency funds or instruct employees to authorise payments. Business email compromise has evolved to incorporate deepfake video: executives are now impersonated in fake video calls directing staff to transfer funds or share credentials. In 2024, a Hong Kong firm lost NZ$40 million after employees were convinced by a deepfake video call impersonating their CFO and other senior staff. Investment fraud uses deepfake videos of celebrities and politicians to promote fraudulent schemes that strip savings from victims who trust what they see.

Globally, AI-enabled fraud is estimated to reach NZ$65 billion in annual losses by 2027. New Zealand is not insulated from this. The country’s small, high-trust social environment in which people are accustomed to acting on a familiar voice creates vulnerability to voice cloning scams. Pacific, Māori, and rural communities, where tight extended family networks mean that a call from a ‘whānau member’ carries immediate credibility, are acutely exposed.

The Bill’s scope confined to intimate visual recordings leaves this entire category of harm unaddressed. The Committee should recommend that the Government urgently consider whether the Crimes Act, the Fair Trading Act, Financial Markets Conduct Act and the Copyright Act require amendment, and whether Netsafe’s mandate should extend to this harm vector.

3. Political Deepfakes and the Threat to Democracy

The Bill does not address political deepfakes. This is its most significant structural gap from a societal perspective. The ability to fabricate convincing audio and video of politicians, candidates, and public figures saying things they never said poses an existential threat to the integrity of democratic discourse.

In the 2024 United States presidential primary, deepfake robocalls impersonating President Biden were distributed to Democratic voters in New Hampshire directing them not to vote. In Australia’s 2025 federal election, research indicated many voters struggled to identify. Australian Greens Senator David Shoebridge warned explicitly that New Zealand should watch what happened there and prepare. The 2023 New Zealand general election saw early use of AI-generated political content, and experts warned of a ‘shit-flood’ disinformation environment in which deepfakes give ‘duplicitous politicians wide air cover for deniability’.

There are currently no laws in New Zealand explicitly prohibiting political deepfakes. Electoral law, defamation law, and broadcasting standards are all poorly equipped to address fabricated audio-visual content that is designed to circulate at speed on social media before any correction can be made. The damage a political deepfake causes is rarely reversible: content that has been seen cannot be unseen, and corrections rarely reach the same audience as the original fabrication.

The consequences for Māori political participation are particularly acute. Māori leaders, advocates, and politicians are public figures whose voices and likenesses are available in digital form. Fabricated content placing extreme, offensive, or divisive statements in their mouths could cause rapid and lasting damage to their political standing and to community relationships. Māori political representation is hard won; deepfake technology is a tool that could be weaponised to undermine it.

I recommend the Committee note that this Bill does not address political deepfakes and direct the Government to introduce dedicated electoral deepfake legislation urgently. At minimum, the Electoral Act should be reviewed for amendment to require disclosure of AI-generated content and to prohibit deepfake fabrications of political candidates and public officials.

4. Platform Accountability and Removal Obligations

The Bill appropriately extends court-ordered takedown and suppression mechanisms to deepfake intimate imagery. However, the current removal framework requires court orders that take weeks or months to obtain. For victims of image-based abuse, speed of removal is critical to limiting harm.

Comparable international frameworks impose platform obligations to act far more quickly. The United States’ Take It Down Act proposes/establishes rapid removal expectations (e.g. 48 hours) Australia’s Online Safety Act imposes systemic risk management and harm prevention obligations on platforms. The EU’s Digital Services Act places a general duty of care on platforms including requirements for transparency and risk management.

I recommend the Committee consider amendments requiring major digital platforms operating in New Zealand to establish expedited removal processes for non-consensual intimate deepfakes, with transparency reporting on removal volumes and timelines. For Māori and Pacific communities where viral spread through networks can cause rapid harm, the 48-hour standard observed internationally should be considered a minimum.

5. Equity of Access to Remedies

Expanding criminal and civil pathways on paper is necessary but insufficient if marginalised communities cannot access those pathways in practice.

A Bill that creates remedies accessible primarily to those with resources and digital literacy replicates systemic inequity in new technological clothing.

I recommend the Committee direct that the Bill’s implementation be accompanied by:

• Targeted resourcing for Māori-led and Pacific-led digital harm support services, funded through the Ministry of Justice’s Treaty obligations;
• Culturally appropriate victim support pathways for LGBTQ+, disabled, and rural communities, developed with those communities;
• A requirement that Netsafe report disaggregated data on complainant ethnicity, disability status, gender identity, and geographic location, enabling monitoring of equitable access to remedies;
• Accessibility commitments ensuring all complaint, reporting, and removal pathways meet New Zealand accessibility standards.

6. Summary of Recommendations

The Committee should recommend the following amendments and accompanying actions:

1. Equity impact assessment: Require a formal equity impact assessment of the Bill’s implementation, with specific analysis of access barriers for Māori, Pacific, LGBTQ+, disabled, rural, and low-income communities.
2. Vulnerable community support: Mandate targeted resourcing for Māori-led and Pacific-led digital harm support services; culturally appropriate pathways for LGBTQ+ and disabled communities; and disaggregated ethnicity, disability, and geographic reporting from Netsafe.
3. Expedited platform removal: Amend the Bill to establish platform obligations for expedited removal of non-consensual intimate deepfakes within 48 hours, consistent with comparable international frameworks.
4. Financial fraud: Recommend that the Government urgently review the Crimes Act, Fair Trading Act, and Financial Markets Conduct Act to address deepfake-enabled financial fraud, including voice cloning scams disproportionately targeting older, rural, and low-income New Zealanders.
5. Copyright Act.: Recommend that the Government urgently review the Copyright Act to include the right of an individual to have copyright to their own image and biometrics.
6. Political deepfakes and democracy: Note that the Bill does not address political deepfakes and direct the Government to introduce dedicated electoral deepfake legislation urgently, drawing on international models from South Korea and Singapore.
7. Future scope: Note in the Committee’s report the identified gaps regarding cultural and ceremonial identity, political deepfakes, financial fraud, and deceased persons, and recommend the Government return to those gaps through further reform.

8. Conclusion

The Deepfake Digital Harm and Exploitation Bill addresses a genuine and urgent gap in New Zealand law. Image-based abuse enabled by AI causes devastating harm to individuals, their families, and their communities and the law must keep pace.

But deepfake technology causes harm well beyond the intimate imagery this Bill addresses. It enables financial fraud that strips savings from elderly and vulnerable New Zealanders. It enables political manipulation that threatens the integrity of democratic participation. And it causes collective as well as individual harm to Māori, Pacific, to the safety of LGBTQ+ individuals whose identity may be forcibly disclosed, and to those in rural, disabled, and low-income communities who face the same harms but cannot access the same remedies.
The amendments and accompanying actions I have recommended are the difference between legislation that helps some people and legislation that works for everyone.

 

In conclusion

Every story that goes untold because mainstream media does not look for it, because shame keeps people silent, or because there is no faith that the system will respond, is a story that lets government and platforms continue operating without consequence. Fixing the law so the next Ario does not have to wait months for a court order, and so her community’s harm is recognised as harm at all, is the rest of it.

The Deepfake Digital Harm and Exploitation Bill is a necessary and welcome step, and it deserves to pass. It is also important that the Bill is expanded to protect other forms of deep fake harms.