top of page
  • HEIA

Trends in Harmful Inaccurate Information in New Zealand

A HEIA Snapshot Report December 2023 – January 2024


Download this report in PDF format:

HEIA_Snapshot_240420_Trends_in_Harmful_Inaccurate_Info_NZ
.pdf
Download PDF • 1.60MB



Introduction

This report is the third in a series of snapshot reports designed to understand themes and trends in harmful incorrect information in New Zealand. This instalment focuses on how these themes and trends developed between late November 2023 and early February 2024.

As discussed in a recent report we use the phrase Harmful Incorrect Information (HII) to refer to the false ideas, beliefs and claims prevalent in New Zealand which are capable of causing a range of harms to individuals, communities and institutions. HII includes both misinformation and disinformation (false information which is spread unintentionally and deliberately, respectively) and can also include other components such as hate speech or incitement of violence involving false information. The spread of HII can erode trust, cohesion, and the health, education and wellbeing of individuals and groups.

Our first snapshot report, published in October 2023, identified four main themes in HII in New Zealand: Health and Covid-19; Globalist Conspiracies; Politics (Distrust and Withdrawal); and Social Anxiety.

Our second snapshot added discussion of the Israel Hamas conflict to our analysis. We reported that access to reliable information about the conflict was being impacted by the ‘fog of war’, where facts on the ground are unable to be easily verified and in many cases have been manipulated to appeal to audiences’ biases.


Global Context

2024 has commenced with misinformation and disinformation listed as ‘the most severe short-term risk the world faces’ in the World Economic Forum 2024 Global Risks Report.[1] With the increasing risk of global conflict, a year of elections vulnerable to being targeted, and rapidly emerging technology, HII poses an evolving threat.

'Misinformation and Disinformation are the most severe short-term risk the world faces' World Economic Forum January 2024

Rising global conflict Disinformation continues to be used as a key weapon in the Russia-Ukraine conflict, and its use in the Israel-Hamas conflict has continued to bring increasing awareness of the threat posed by the manipulation of the information environment. Rising escalations in the region have not been immune and there is evidence that countries like Russia are exploiting the information fog to disrupt the West’s role in the region[2]. HII can exacerbate existing tensions and undermine efforts to resolve conflicts through the sharing of false and misleading information. HII, often consumed as a result of algorithms, can also heighten existing political and cultural biases amongst the public, heightening societal polarisation. 

A year of elections 2024 is the biggest election year in history, with over two billion people eligible to cast votes around the world, including in the USA and India. HII, including that promoted using artificial intelligence (AI), is a threat to free and fair elections with interference and influence in the democratic process. Even in New Zealand, where the information ecosystem is still robust and trust in government remains high, the 2023 general election showed indications of HII impacting the views of some voters but not the electoral process overall as evidenced in our previous report.

Emerging role of AI in disinformation campaigns Generative AI that can create new text, images and other media is increasingly becoming one of the most disruptive technologies in existence. The potential for malicious use will test all forthcoming democratic elections, not least in the United States as malign actors ramp up their methods with targeted campaigns and flooding the information environment. Examples are already being widely circulated of false endorsements for candidates or content that undermines a candidate’s credibility including President Joe Biden[3]. One advocacy group in the US has stated “the political deepfake moment is here”, highlighting growing concerns about the regulatory gaps around AI accountability and transparency.[4]

'The political deepfake moment is here' 

In addition, the increase of non consensual deepfakes has been brought to the forefront with well known public personalities such as singer Taylor Swift. This highly publicised activity has helped to expose how the internet is being flooded with AI generated pornography, often fuelled by misogyny, and has prompted a US bill to allow civil penalty against perpetrators.[5] 

Climate change denial Recent research published by the Center for Countering Digital Hate (CCDH) shows how climate change denialism is changing from the outright rejection of climate change as an existing phenomenon to promote campaigns that undermine potential solutions and delay political actions.[6] Researchers and advocates are calling for a shift of focus away from countering climate change denialism to increased transparency on the need for suitable solutions as a result.


[1] World Economic Forum (2024). These are the 3 biggest emerging risks the world is facing. Weforum.org. Retrieved from https://www.weforum.org/agenda/2024/01/ai-disinformation-global-risks/

[3] Yerushalmy, J. (2023, 23 Feb). AI deepfakes come of age as billions prepare to vote in a bumper year of elections. The Guardian. Retrieved from https://www.theguardian.com/world/2024/feb/23/ai-deepfakes-come-of-age-as-billions-prepare-to-vote-in-a-bumper-year-of-elections  

[4] Elliott, V., Makena, K. (2024, 23 Jan). The Biden Deepfake Robocall Is Only the Beginning. Wired. Retrieved from https://www.wired.com/story/biden-robocall-deepfake-danger/

[5] Durbin, R. (2024). The DEFIANCE Act of 2024. Retrieved from https://www.durbin.senate.gov/imo/media/doc/defiance_act_of_2024.pdf (Accessed 28 Feb 2024)

[6] Centre for Countering Digital Hate. (2024). The New Climate Denial. Retrieved from https://counterhate.com/research/new-climate-denial/ 



Methodology

This report collects and analyses anonymous New Zealand-based posting data from publicly available platforms where harmful inaccurate information (HII) is prevalent. Platforms include Facebook, Instagram, Telegram, Reddit. We do not capture data from all online platforms, and we therefore do not claim to provide a complete picture of HII in New Zealand. The purpose of HEIA’s snapshot reports is to provide a measured indication of trends in conversation, why they might occur, and where.

For the reporting period 26 November 2023 - 11 February 2024 our total dataset comprises 892,600 New Zealand-based posts / online entries. Of this dataset we identified approximately 1,856 posts which are confirmed HII.[7] This is 0.2% of our dataset of New Zealand-based posting. This total is made up of the following number of posts for each theme: Health and COVID-19 (891 posts), Globalist Conspiracies (1074 posts), Politics/Distrust and Withdrawal (200 posts), Social Anxiety (185 posts). A further 76 posts confirmed as HII did not align with the above categories and were marked as miscellaneous.

In contrast to previous reporting, in this report we have used a proprietary Large Language Model (LLM) to detect HII from a very large dataset of online posting. This is an improvement to the method to allow more accuracy of capture when paired with manual verification.

Subsequent manual verification of data first identified as HII by the LLM by a team of researchers removed posting which did not meet our threshold of inaccuracy and harm. Once posting data had been confirmed as HII, the posts were then categorised by key theme and subtheme. Often posts were attributed to multiple categories, given the nature of certain conspiracy narratives. The few posts that did not align with a key theme were attributed to a miscellaneous category.

Because our method of filtering data has changed to capture only HII (and not HII-related content), our results are more conservative (i.e. identifying a lower volume of HII) compared to previous reporting. This enhanced method using an LLM designed to detect and analyse HII / disinformation has greater utility to identify emerging themes in the future. However, as discussed above, because this tool is in the early stages of development, substantial manual checking of the data is still required and was conducted for this report.

All data is anonymised on collection and no personal information is intentionally collected or used in HEIA’s research. In cases where users have included personal information in the content of their posts, it may be collected unintentionally. Any personal information identified is deleted and we do not provide any such information or other data to third parties and do not use it in our analysis.

[7] We make a clear distinction between HII and what could simply be deemed ‘(incorrect) statements of opinion’. To be categorised as HII a post must not only be inaccurate (as many misinformed but benign opinions are), but also liable to cause genuine harm.


Key Themes

From a previous qualitative and quantitative review of the majority of mainstream and fringe online platforms we identified four main themes in harmful inaccurate information (HII) in New Zealand.

  • Theme 1, Health and COVID-19 is HII related to distrust in the official health narratives and initiatives of the New Zealand Government and international health organisations.

  • Theme 2, Globalist Conspiracies is HII infused with the belief that world events, including in New Zealand, are being controlled by an elite cabal of ‘globalists’ intent on creating a ‘New World Order’. 


  • Theme 3, Politics (Distrust and Withdrawal) is HII which maintains the New Zealand Government is undemocratic or cannot lawfully control individuals. Elections are rigged, and the government is authoritarian and / or tyrannical.

  • Theme 4, Social Anxiety is HII associated with a belief that society is moving in the wrong direction. It often includes suspicion of any attempt to support marginalised communities.

In the following section we discuss each of these themes in turn and the volume of posting associated with each.

Figure 2 - Number of Posts per fortnight by thematic group (1 Jan - 26 Nov 23)
Figure 1: Number of daily posts by theme (26 Nov 23 - 11 Feb 24)

Theme 1: Health and COVID-19

Figure 2 - Number of Posts per fortnight by thematic group (1 Jan - 26 Nov 23)
Figure 2: Number of daily posts - health and COVID-19 (26 Nov 23 - 11 Feb 24)

Though the numbers remain low, comparative to our previous report, HII relating to health and COVID-19 rose in volume in early December. This is an interesting development, given we had observed a gradual decline in similar discussion over consecutive months. Previously we had contended that topical fatigue and the passage of time since the height of the pandemic had driven this trend. However, whilst noticeable, the uptick in conversation still does not represent significant volume at this stage. Two particular topics of conversation appear to be driving the trend line. First, in early December Te Whatu Ora experienced a data breach when one of its employees, an activist, leaked the private vaccination data of thousands of New Zealanders. Online HII can be seen to surge at this point in time, with numerous posts addressing the activist’s arrest with accusations of government tyranny and/or the unseen hand of the deep state: “NZGovt Health is run from offshore by the WWW Criminati. Watch the buggers throw [the activist] under the bus in court” (Facebook, 27/01/24). Amongst the vaccine sceptic community, the activist has been hailed as a hero, said to have revealed evidence of a government-orchestrated cover up of vaccine-related injuries and deaths. He is ascribed whistleblower status by the community, inferring that he has duly brought the proposed cover up to the public’s attention. The content of much of this discussion illustrates the emotion that, for many, still accompanies memories of pandemic experiences. A related ‘subplot’ within the above narrative is that leaders of the so-called freedom movement are being systematically targeted by authorities for releasing information deemed to be in the public interest. A prominent activist, for example, went into hiding in an ‘undisclosed location’ following the data breach. There is a sense in many of the posts that this incident has huge gravity, a landmark moment for the freedom movement, with the eyes of the world on New Zealand.

A second common topic has been discussion of the forthcoming Royal Commission of Inquiry into Lessons Learned from Aotearoa New Zealand’s Response to COVID-19. Claims of a whitewash, aimed to discredit the Inquiry before it begins, were notable in our data: “rather than looking to find fault or assign blame”, they “are focused on capturing the lessons learned from the pandemic” so New Zealand is prepared for the future "pandemics" the WHO will roll out …Basically, as we stated from the outset this will be a whitewash feel good nothing-burger. A CONQUIRY!” (Telegram, 08/02/24). This post stands out as an example of malinformation, manipulating the statements of the Inquiry by inserting them into an imagined draconian context.

Both topics – the data breach and Inquiry - illustrate that COVID-19, and particularly vaccination mandates, remain highly significant for the broader anti-government environment. Triggers such as these will likely ensure that COVID-19 remains a key HII theme in 2024.


Theme 2: Globalist Conspiracies

Figure 4: Number of daily posts - health and COVID-19 (1 Oct - 25 Nov 23)
Figure 3: Number of daily posts - Globalist Conspiracies (26 Nov 23 - 11 Feb 24)

Globalist Conspiracies-related HII remains relatively constant through the reporting period, driven largely by the Elite Control discussional subcategory.

'Discussion relating to Elite Control picked up again in late November and through December'

In our previous report, which analysed data through October-November 2023, we observed how HII relating to Elite Control was experiencing a steep decline. We hypothesised that this lull would be temporary, due to the time it took for the coalition government to form following the general election. Our latest data affirms this, with online discussion relating to Elite Control picking up again in late November and through December. Examples of such discussion continue to implicate NZ’s government as a puppet of the deep state: “[prominent politicians] are WEF puppets” (Facebook, 26.11.23); “the law works for blackrock the mps are owned by blackrock” (Facebook, 15.12.23). Several posts claim the prime minister is ‘anti-farming’, an accusation woven into the belief that the PM is actively working against domestic interests in favour of a shadowy WEF agenda. Some posts caution that, influenced by a ‘globalist plan’ NZ is following European countries in its treatment of farmers (and that farmers will soon resort to similar measures of protest as seen in Europe). We note that a majority of COVID-19 related HII also includes an ‘elite control’ element, where vaccines are deemed part of a control narrative perpetuated by bodies such as the IMF, WEF, and WHO, in collusion with billionaire philanthropists and New Zealand’s government to enact a depopulation agenda.

A short spike in conversation is observed in the two days following the swearing in of the new government. A second spike in mid-January did not, as we might have expected, reflect the Treaty Principles bill memo which was leaked at the time. Instead the data shows a general surge in unconnected conversation, a broad mix of HII narratives that make up the Globalist Conspiracies category. The largest data spike, observed in late January, reflects similarly varied discussion, with no one theme driving the trend. Certain topics labelled in our data as ‘miscellaneous’ can be seen to rise slightly at this point. Analysis of the posts shows these miscellaneous themes include that the CIA is funding Ukrainian nazi militias, that Ukrainian authorities are running a child trafficking ring, and that a prominent member of the royal family is a homosexual.

It should be noted that this is all occurring at fairly low volume, however, and that it does not reach the levels we observed in October, leading into the general election. As with our previous reporting, our three other subcategories under the Globalist Conspiracies theme - relating to Financial HII, Natural Events, and Tech and Surveillance - continue to exist at very low levels.


Theme 3: Politics (Distrust and Withdrawal)

Figure 5: Number of posts per fortnight – Globalist Conspiracies (1 Jan - 26 Nov 23)
Figure 4: Number of daily posts - Politics (Distrust and Withdrawal) (26 Nov 23 - 11 Feb 24)

Our reporting in November 2023 noted that, following the general election, the Politics (Distrust and Withdrawal) theme declined precipitously across all subcategories (Stolen Election, Tyranny, Sovereign Citizen). It is reassuring that the trend line has continued to fall with the coalition government now in its third month of office.

Most prominently, the Stolen Election theme has vastly decreased in volume compared to related conversation leading into the general election. Moreover, the posts that do discuss stolen or rigged elections are all in the context of the US, with no posts referring to the recent NZ election. A single notable spike in Tyranny-related discussion seemingly reflects accusations of an ‘authoritarian’ like response to the Te Whatu Ora data breach, together with related discussion of what the leak was thought to evince; a tyrannical government cover up of vaccine-related deaths - part of its assumed plan for population control.

'The Stolen Election theme has vastly decreased in volume compared to related conversation leading into the general election'

We were somewhat surprised to see a lack of engagement with two prominent news stories in December 2023 and January 2024; in the first instance, public sector usage of te reo Māori, and second, the leaked Treaty Principles bill memo. Where we had expected to see an uptick in Sovereign Citizen related posting, we instead observe the Sovereign Citizen subcategory all but flat line through this period. It is important to note, however, given that interpretations of the Treaty and Māori sovereignty remain contentious, that we do not class all related discussion in our dataset as HII, since we cannot clearly define it as ‘inaccurate information’.


Theme 4: Social Anxiety

Figure 10: Number of daily posts – Politics (Distrust and Withdraw) - Subthemes (1 Oct - 25 Nov 23)
Figure 5: Number of daily posts - Social Anxiety (26 Nov 23 - 11 Feb 24)

Like the Politics trend, HII relating to Social Anxiety has waned in our November-February data. Indeed, of our four primary HII themes, Social Anxiety has seen the most dramatic drop off in the first two months of 2024 .

'A small but noticeable spike in the ‘Culture Wars’ subtheme in January appears to be driven by a single Telegram campaign'

A small but noticeable spike in the ‘Culture Wars’ subtheme in January appears to be driven by a single Telegram campaign (likely the same user posting on different channels). The posts largely relate to a claimed exposé of Hollywood celebrity pedophilia. Across the reporting period we were somewhat surprised to see a lack of HII relating to Gender, Misogyny, and LGBTQIA+ where previously there had been relatively ongoing conversation. One explanation may be the relative absence of the media spotlight on these topics in recent weeks; in past reports we noted that the Gender, Misogyny and Anti-LGBTQIA+ trend line often reflects the media cycle.

Of further note is that the Anti Co-Governance subcategory has continued to flatline. One explanation for this may be the current government’s stance on Three Waters legislation, a contentious issue which had previously been driving HII relating to co-governance. Given that we have observed a declining trend line over four consecutive months we may conclude that Anti Co-Governance is no longer a key focus of HII in Aotearoa. We will review its inclusion based on data ahead of our next report.


Conclusions

Continued low volume of HII-content as a proportion of New Zealand-based posting: We continue to see that HII content only comprises a very small proportion of our overall dataset of New Zealand-based posting. We caveat once again, however, that this report has not included analysis of the Israel Hamas conflict, a topic that has sparked much online discussion and which would likely increase the proportion of HII as a percentage of overall posting.

Rise in Health & COVID-19 HII: Prior to the current reporting period we noted a decline in COVID-19 related HII over consecutive months. From late November we observed a trend shift. Two narratives are identified as driving conversation as a likely  result of astroturfing[8] in response to media activity, a known disinformation tactic: the Te Whatu Ora data breach[9], and the Royal Commission of Inquiry into the pandemic response. These spikes indicate how high profile offline events can drive surges in online HII, which policy makers should be prepared for. However, we note that overall thematic discussion remains at a low level.

'Ideas of globalist conspiracy have moved seamlessly from implicating the previous government to targeting the new one'

Confirmation of transfer of Globalist Conspiracy narratives to new government: Ideas of globalist conspiracy have moved seamlessly from implicating the previous government to targeting the new one. We began to observe this trend in our previous report. We contend that the transnational nature of the information environment results in a local conspiratorial milieu that is affected as much (if not more) by international trends as by domestic politics.

Decline in HII relating to Politics (Distrust & Withdrawal), and Social Anxiety: We observed a decline in almost all subcategories under our Politics and Social Anxiety themes. Where HII tends to follow the media cycle we contend that subcategories like Gender, Misogyny, and Anti-LGBTQIA+ have not been under the media spotlight during the reporting period. Subcategories which were beginning to gain traction in the lead up to the general election (Anti Co-Governance, Sovereign Citizen, Stolen Election) have dissipated since the appointment of the new government.

Recurring miscellaneous themes: Our data displayed a number of topics of discussion which did not align with our key categories. The most common of these topics was the war in Ukraine, with recurrent HII narratives including that the CIA is funding Nazi militant groups, and that the Ukrainian government is engaged in child trafficking. Separately, we noted an ongoing campaign to seed the idea that a prominent member of the royal family is gay. (This royal is often implicated in the globalist conspiracy/new world order narrative). We note that the above miscellaneous themes were observed at a lower rate than other themes analysed. 

Recurring miscellaneous themes (non-HII): The Large Language Model used for initial data collection captured several posts which were later confirmed not to be HII. The most common posts in our data which were not deemed HII were commentary on the Israel Hamas conflict. We note with interest that the second most popular topic was heated commentary on the NZ government’s planned legislation of tobacco sales.

We will continue to monitor levels of HII in New Zealand in our next report. Given the low volume of posts in several thematic categories we may review how we categorise posts by topic in future reporting.


References

Centre for Countering Digital Hate. (2024). The New Climate Denial. Retrieved from https://counterhate.com/research/new-climate-denial/

Chan, Jovy. (2022). Online astroturfing: A problem beyond disinformation. Retrieved from https://journals.sagepub.com/doi/full/10.1177/01914537221108467.

Durbin, R. (2024). The DEFIANCE Act of 2024. Retrieved from https://www.durbin.senate.gov/imo/media/doc/defiance_act_of_2024.pdf (Accessed 28 Feb 2024)

Elliott, V., Makena, K. (2024, 23 Jan). The Biden Deepfake Robocall Is Only the Beginning. Wired. Retrieved from https://www.wired.com/story/biden-robocall-deepfake-danger/

Institute for Strategic Dialogue. (2023). Capitalising on crisis: Russia, China and Iran use X to exploit Israel-Hamas information chaos. Retrieved from https://www.isdglobal.org/digital_dispatches/capitalising-on-crisis-russia-china-and-iran-use-x-to-exploit-israel-hamas-information-chaos/

World Economic Forum (2024). These are the 3 biggest emerging risks the world is facing. Weforum.org. Retrieved from https://www.weforum.org/agenda/2024/01/ai-disinformation-global-risks/

Yerushalmy, J. (2023, 23 Feb). AI deepfakes come of age as billions prepare to vote in a bumper year of elections. The Guardian. Retrieved from https://www.theguardian.com/world/2024/feb/23/ai-deepfakes-come-of-age-as-billions-prepare-to-vote-in-a-bumper-year-of-elections


Disclaimer

This report was commissioned by the Department of the Prime Minister and Cabinet (DPMC) and prepared independently by HEIA. The report does not represent the views of DPMC or the New Zealand Government.

At DPMC’s request, information that could identify any individual person is redacted to protect privacy.


For enquires regarding this report please contact Dr. Chris Wilson at chris@heiaglobal.com







Download this report in PDF format:

HEIA_Snapshot_240420_Trends_in_Harmful_Inaccurate_Info_NZ
.pdf
Download PDF • 1.60MB

Comentários


bottom of page