Truth In Political Campaigns: Investigating Artificial Intelligence Prevalence Across the United States, India, and Australia, 2019-2026

Authors: David Guo, Patricia Larasgita, Tuan Hung Mai, Vanshika Saini and Isabella Yu

---

EXECUTIVE SUMMARY

Digital spheres have become increasingly powerful tools in global political campaigns over the past decade and within this trend, the impacts of artificial intelligence (‘AI’) have materialised concretely and consequentially. This report is a comparative study between the United States, India, and Australia — chosen to provide samples across populations and scrutiny on their politics — and traces the use of AI in political campaigns across the decade. Our findings indicate a similar trend between the three countries, notably within the past three years, as there has been a prominent increase in AI’s relevance, whether via deepfakes, algorithm manipulation, or bots. AI’s prevalence in Australian political campaigns occurs noticeably later but the risks of hacking and misinformation faced in India and the United States foreshadow the challenges it will confront (Worthington 2019, Hui 2025). 

Contextually, the Australian populace’s sources of news have shifted from traditional broadcast media to social media, rising from 18% in 2019 to 26% in 2026 (Park et al. 2025, 75). This reflects global trends, including in the US and India. The US has observed social media overtaking traditional news media altogether and India has seen a particularly sharp rise amongst younger populations (Newman et al. 2025, 119). Australia has not engaged in tangible legislative reform in this space despite awareness of the technology's capabilities and warnings from the Australian Electoral Commission over its limited capability to detect and respond to AI-generated posts (Gould 2024). As such, AI’s impact on the larger population samples of India and the US would be largely undeterred in Australia’s current legislative state.

---

METHODOLOGY

This report  analysed examples of deepfake use in political advertising across three jurisdictions: Australia, the United States, and India. While deepfake use in political communication exists in other jurisdictions, we have not pursued study of those in this report.  As such, while this report provides a useful snapshot of how deepfake technology has been used by political advertisers in these jurisdictions, it does not seek to establish a comprehensive summary of all use of deepfake technology by political advertisers in these jurisdictions. Rather, it aims to provide Australian policymakers with a snapshot of how deepfake technology has been used by political advertisers in these jurisdictions that may be used in future regulatory contexts. 

Due to varying levels of data available in each of these jurisdictions, different approaches were used to research deepfake technology use in political contexts across each jurisdiction. As this report is a summary of deepfake technology usage by political advertisers, this does not affect the quality of the research below. However, a comprehensive review of this subject would in the future require an aligned approach to research.  

The table below sets out the methodology applied across each jurisdiction:

Table 1.1: Research methodology across each jurisdiction.

---

TERMINOLOGY

1: DEEPFAKES

A deepfake refers to a specific kind of synthetic media where a person in an image or video is swapped with another person's likeness (Somers 2020). It is created by artificial intelligence systems based on various media elements. As technology continues to develop, deep-fake production has become more accessible and capable of generating increasingly hyper-realistic outputs over time.  In politics, deepfakes can serve several functions. For example, they can be used damage the reputation of opponents, mobilise the political base, undermine trust in the election process, shape policy discourses or promote polarising rhetoric (Weiner 2024).  

As an example, deepfake can be used to persuade public opinions shape public perceptions of political candidates by minimising perceived flaws. Advances in deepfake technology and its growing affordability have increased public concern about its use in political campaigns. During the 2024 US presidential election, a survey showed around 77% of respondents had encountered deepfake content related to political campaigns, while 74.7% of respondents expressed concern about its potential impact in the upcoming election. (Łabuz and Nehring 2024). 

Although deepfakes have strong persuasive potential, developing an effective detection system that can keep pace with their rapid evolution remains challenging (Weiner 2024). Despite their potential risks, deepfakes have not historically played a significant role in shaping election results. Instead, their presence has contributed to growing public scepticism, prompting individuals to question the authenticity of political media and to actively verify the information they receive. Nevertheless, the amount of deepfake content shared by political actors is growing, and is contributing to worsening information environments (Łabuz and Nehring 2024).   

2: BOT ACCOUNTS AND ALGORITHM MANIPULATION

A bot is a user account that interacts in automated ways (Howard, Woolley, and Calo 2018). Bots are often designed to perform certain functions such as sharing information at specific times, creating content based on human input, or interacting with human users. Some bot accounts are useful tools such as chatbots that respond to common questions of users, or knowbots that are designed to automatically retrieve some useful information from the internet (Dehghan et al. 2023).

In political contexts, bot accounts often programmed to mimic human behaviour, manipulate public opinion, and amplify divisive narratives; shaping perceptions surrounding political movements or issues. These tactics pose challenges to differentiate authentic discourse from orchestrated campaigns, as genuine public sentiment and artificially manufactured narratives are becoming harder to distinguish in the information environment (Feldman and Nahmias 2025). 

Meanwhile, algorithms are tools that curate content on digital platforms by predicting what will keep audiences engaged. This process encourages users to access information based primarily on their personal preferences (Fu and Sun 2024). As a result, it can reinforce echo chambers on social media by limiting users’ exposure to diverse perspectives, thereby further destabilising democratic discourse (Australian Government n.d.).

In the political context, algorithm manipulation occurs when political actors intentionally exploit these engagement-driven systems to increase the visibility of their preferred content. Bots are often used as a key mechanism in this process, as they are deployed to artificially amplify content and trick algorithms into promoting it (Feldman and Nahmias 2025).

3: AI HALLUCINATIONS

The Australian Government website describes AI hallucinations as “output flaws" including "factually incorrect, nonsensical, or misleading information” (Australian Government n.d.). In Chapter 2 of its Interim Report on the impacts of AI on democracy, the Senate Select Committee on Adopting Artificial Intelligence warned that where the operational processes behind AI computations remains opaque, such lack of understanding may generate not only AI hallucinations, but “inconspicuous discrepancies and discriminatory biases” capable of undermining the fairness of elections if implemented without considerate human oversight (Senate Select Committee on Adopting Artificial Intelligence 2024). As discussed below, the South Australian algal bloom in September 2025 illustrates how hallucinations can enter political debate when politicians rely on AI-generated scientific findings without independent human verification. The consequences of such events extend far beyond a one-time political embarrassment — the use of hallucinated information in political contexts can erode public trust in both media and government, particularly when such actors are perceived as making no genuine effort to research and distinguish their materials at hand.

---

CASE STUDY 1: UNITED STATES

This section examines the systemic shift in political communication from the 2016 cycle to the present day (2026). It conceptualizes the integration of artificial intelligence (AI) and synthetic media (deepfakes) not merely as a technological evolution, but as a fundamental restructuring of the marketplace of ideas into a fractured information ecosystem. The seminal literature on AI in elections began with the study of Computational Propaganda (Woolley and Howard 2016). This framework describes the use of algorithms, automation, and human-curated bots to manipulate public opinion. In the 2016 context, this manifested through Algorithmic Microtargeting — the use of machine learning to "profile" voters and deliver personalised, often inflammatory, messaging.

As the technology progressed, scholars introduced the Liar's Dividend (Chesney and Citron 2019). This theory argues that the proliferation of deepfakes allows political actors to evade accountability by falsely claiming that authentic, damaging evidence is AI-generated. Recent experimental studies involving over 15,000 participants confirmed that this strategy is increasingly effective, as "crying wolf" over misinformation can bolster support for a politician even when the evidence against them is real (Schiff et al 2024).

Ultimately, these tactics converge into 'epistemic fatigue’ as unlike traditional propaganda that aims to persuade, modern AI-driven disinformation often aims to overwhelm. By flooding the social media landscape with "banal dystopias" of conflicting synthetic realities, malicious actors induce a state of cognitive exhaustion where voters cease to believe in the possibility of objective truth (Schick 2020; Robins-Early 2023).

CHRONOLOGICAL ANALYSIS OF AI AND DEEPFAKE IMPLEMENTATIONS (2019–2026)

The evolution of AI in US elections follows a clear trajectory: from invisible algorithms to visible, although misleading, realities. From 2019–2022, an evident transition into deepfakes is visible. 

In 2019, a low-tech, slowed-down “cheapfake” video of US House Speaker Nancy Pelosi was circulated on Facebook. The viral manipulation of Nancy Pelosi’s speech served a seminal case study for how "shallow" AI tools, used for audio manipulation, could destabilize political trust. The video was paired with distorted audio, making her appear intoxicated and slurring her words. It was widely shared by users across social media platforms, including by Donald Trump’s attorney, Rudolph Giuliani (Weiner 2024).

In 2020, an Obama/Peele deepfake was produced by BuzzFeed and Jordan Peele. This video used generative adversarial networks (GANs) to create a flawless likeness of Barack Obama. Though meant as a warning, it functioned as a "dual-use" technology proof-of-concept that lowered the psychological barrier for political forgery (Chesney and Citron 2019; Schick 2020).

During the 2020 cycle, Facebook removed networks of accounts using AI-generated "GAN faces" for profile pictures. These synthetic humans were used to manage "patriotic" groups, making foreign-led influence operations look like domestic grassroots movements (Wahl-Jorgensen and Carlson 2021).

In 2023–2024, the use of generative AI marked a notable shift in its widespread use and its growing influence on media and political discourse. Following Joe Biden’s re-election announcement, the Republican National Committee released an AI-generated advertisement depicting a fictional war-torn America. This marked the first time a major political party used generative AI for official campaign materials (Robins-Early 2023; Al Jazeera 2024).

In April 2023, an AI-generated video was circulated depicting Hillary Clinton praising Florida Governor Ron DeSantis and endorsing him as a Republican presidential candidate. Although the original creator of the video could not be identified, it was widely disseminated by Trump supporters (Łabuz and Nehring 2024).

In June 2023, Ron DeSantis’s campaign used AI to generate photorealistic images of Donald Trump embracing Dr. Anthony Fauci. These images were seamlessly blended with real photographs, forcing voters to perform "forensic analysis" on social media posts just to understand basic campaign facts (Łabuz and Nehring 2024).

In 2023, An AI-generated image depicting Donald Trump dancing with a 13-year-old girl circulated online. Trump’s presidential campaign accused Ron DeSantis's campaign team of creating the image. DeSantis’s campaign team denied the allegation (Łabuz and Nehring 2024; Associated Press 2023).

In 2023, a television advertisement featured Donald Trump’s voice attacking Iowa Governor Kim Reynolds. However, the audio was not Trump’s actual voice; it was AI-generated and created by a political action committee supporting Ron DeSantis (Łabuz and Nehring 2024; Politico 2023).

In December 2023, an AI-generated video portrayed CNN news anchor Anderson Cooper admitting that Donald Trump had humiliated CNN. The video was shared by Trump and members of his campaign team as an attack on the media outlet (Łabuz and Nehring 2024).

In January 2024, an AI-cloned voice of Joe Biden was used to call thousands of New Hampshire voters, telling them to stay home during the primary. This case is frequently cited by the Partnership on AI (2024) as the most dangerous use of synthetic audio to date, as it targeted the "last mile" of voter behaviour — the act of showing up to the polls (Partnership on AI 2024; Yan et al. 2025). It was AI-generated audio created by Steve Kramer, a political consultant. Kramer later claimed he produced the audio independently, despite collaborating with another political candidate (Feldman and Pappu 2025; The Guardian 2024; CNN 2024).

In 2024, when The Lincoln Project released ads featuring authentic but unflattering clips of Donald Trump, the candidate claimed they were "AI-generated deepfakes." This served as the premier real-world example of the Liar's Dividend, where a politician used the existence of AI to delegitimise truthful reporting (APSA 2024; Youngblood 2026).

In August 2024, Trump shared AI-generated images of Taylor Swift and fans in patriotic attire endorsing him. This highlighted the use of AI to fabricate "social proof," where synthetic endorsements create a false sense of momentum and consensus (Youngblood 2026; Al Jazeera 2024).

In October 2024, Microsoft-monitored Russian actors released deepfake videos depicting Kamala Harris making derogatory comments and engaging in illegal poaching in Zambia. These videos gained millions of views on X within 24 hours, showing how foreign adversaries used AI to "hyper-personalise" character assassination (Al Jazeera 2024).

In 2024, North Carolina gubernatorial candidate Mark Robinson dismissed scandalous but authentic reporting as AI-driven fabrication. This exemplified how local and state elections became testing grounds for the same AI-denial strategies used at the federal level (Al Jazeera 2024).

---

CASE STUDY 2: INDIA

The overall trend in Indian elections is that deepfakes are commonly used by both official and anonymous users to create AI-generated videos of popular figures to advance their agenda. This includes using the voices and likenesses of deceased politicians and celebrities to promote their candidates or to criticise opponents. Political parties either use “off-the-shelf” technology or commission private firms to create these deepfakes (Rebelo 2024, 8). There is growing concern amongst AI start-up founders about the use of deepfakes in elections. They are afraid that if the use of deepfakes becomes too extreme, it might result in a “blanket ban” of AI usage for election campaigns by the government, which could hinder the industry's technological progress. This led to three AI start-ups, Muonium AI, Dubverse, and Polymath Synthetic Media Solutions, to co-create the Ethical AI Coalition (Rebelo 2024, 8).  

2020:

The first instance of deepfakes used by an official political party on record was in 2020. A politician, Manoj Tiwari, of the BJP party used deepfakes to dub their video into another language in the hope of connecting with voters from other ethnic groups. Ideaz Factory, a political consultancy firm, was commissioned to create this deepfake in 22 languages and 1,600 dialects (Dasilva et al. 2021, 304). The video of Tiwari speaking a Hindi dialect in particular garnered around 15 million views (Łabuz & Strnad 2025, 7).

In 2020, near the Delhi Assembly Elections, two videos circulated on the internet of Manoj Tiwari, the BJP leader in Delhi, attacking an opposition party. The original video was a discussion of the Citizenship Amendment Act. Deepfakes tools were used to manipulate the video (Neyazi et al. 2024, 4).

The use of deepfakes by political parties became more prominent in leading up and during the 2024 Indian general election:

2023:

A representative of the Bharatiya Janata Party (BJP) released a voice recording in which a Dravida Munnetra Kazhagam (DMK) leader accuses his own party of participating in illegal financial transactions. The DMK leader denied this accusation, alluding to possible deepfakes used to create this recording. Forensic work to verify the recording was inconclusive (Labuz & Nehring 2023, 463).

Deepfake technology was used to create videos of Prime Minister Modi singing popular songs. These videos garnered millions of views and appear to have positively influenced his public image by humanising him and fostering a sense of connection with voters (Labuz & Nehring 2023, 463).

2024:

On the 6^th of January, the face of K. Annamalai, the BJP's southern leader, was inserted into a popular meme of a boy known for his “aggressive sales tactics”. This Instagram video insinuates that Annamalai is trying very hard to sell his policy to the people in a state where the BJP is not popular (Rest of World 2024).

On the 7^th of January, The Dravida Munnetra Kazhagam (DMK) party created AI-generated videos of its late leader supporting his son, the current party leader, in the campaign for the state of Tamil Nadu. These videos feature the late leader praising his son at different events (Rest of World 2024).

On the 25^th of January, an AI-manipulated video of H. Vasanthakumar, a late politician, circulated on Instagram through a film account showing him supporting his son’s election effort (Rest of World, 2024).

On the 20^th of February, the Bharatiya Janata Party (BJP) used AI to recreate the voice of Mahendra Kapoor, a beloved late singer, praising Prime Minister Modi and acknowledging his accomplishments (Rest of World 2024).

On the 21^st of February, the Indian National Congress party shared a deepfake video of Prime Minister Modi singing the song “Thief” on its official Instagram account. AI was used to insert the Prime Minister’s voice and face onto the video of the original artist. The post implies corruption within Prime Minister Modi’s government (Rest of World 2024).

On the 24^th of February, the All India Anna Dravida Munnetra Kazhagam (AIADMK) party, like the DMK, uses AI to reconstruct the voice of its beloved late leader, J Jayalalithaa, delivering a speech calling for support in the party's effort to return to power (Rest of World 2024).

On the 1st of March, the BJP distributed an AI-altered video depicting Rahul Gandhi, leader of the Indian National Developmental Inclusive Alliance (INDIA), criticising Mamata Banerjee, another INDIA member. Gandhi's face was inserted in a video of a politician criticising another politician. Both the politician's likeness in the original video and its content were altered (Rest of World 2024).

On the 17^th of April, a video showing a famous actor, Ranveer Singh, criticising Prime Minister Modi, even though the original video showed him praising the Prime Minister. His voice was cloned to alter the video's content (Rest of World 2024).

On the 18^th of April, a member of the BJP party used AI tools for voice cloning and lip-matching to create personalised videos to motivate its campaign workers. AI was used to allow the speaker to address each person's name individually in the personalised video they received (Rest of World 2024).

On the 29^th of April, an AI-generated video of the Indian National Congress’s leader getting sworn in as prime minister, while the voting process has not ended in the election. This video circulated widely amongst supporters of the party (Rest of World 2024).

On the 4^th of May, the Communist Party of India-Marxist (CPI-M) used AI to create a video of their late leader, Buddhadeb Bhattacharjee, giving a speech supporting candidates on the left of the political spectrum and criticising the ruling party (Rest of World 2024). 

On the 17^th of May, an AI-generated video of famous politicians and new anchors at a rally while giving a speech supporting the policy of their opposition. The video was distributed on X, with a caption warning people about the dangers of AI (Rest of World 2024).

---

CASE STUDY 3: AUSTRALIA

One of the first reported examples of AI use in elections came prior to the Queensland state election in 2020, when the Advance Australia lobby group published a deepfake video of then-Premier Annastacia Palaszczuk saying that the state was ‘cooked’ and ‘in massive debt’. The clip amassed approximately one million views, highlighting the substantial reach of the technology even in its early stages of development and that even synthetic media could achieve substantial reach in Australian political discourse (Ray 2021, 988). There are a few other documented instances of artificial intelligence used in political or electoral contexts any earlier than 2023, but from that point, the applications of AI in politics have span across three broad yet overlapping categories of deepfake contents, bot accounts and algorithm manipulation, AI hallucinations and specifically entering official political records, which they improves and upgrades at a generally lowering costs. Artificial intelligence in itself is a neutral technology, yet its political applications by its users are not. Together, these case studies set out below illustrate both the growing normalisation of AI within Australian political environment and the widening gap between existing technology regulatory frameworks and the current pace of technological development. 

Academic literature anticipated these risks in as early as 2020. Paterson and Hanley (2020) framed deepfakes as a sub-component of ‘political warfare’ and a cyber subversion posing a foreseeable and highly destabilising effect to Australian democratic values. Yet, the empirical evidence suggests such trend has not manifested. In their analysis of Twitter activity across the 2013–2019 federal elections, Bruns, Angus and Graham (2021) discovered that only 137 of ~79,000 Twitter accounts (0.17%) met their conservative bot threshold, producing only 0.02% of all election tweets and indicating that bot-driven distortion of Australian federal campaigns was empirically negligible in the pre-generative-AI era. Ray (2021) subsequently flags two distinct categories of harms posted by Artificial Intelligence to Australian elections: direct manipulation of individual voters, and broader erosion of public trust in video as evidence, noting that even swaying 100 voters in a marginal seat can change a federal outcome. More recently, Turner (2025) considered Parliament’s constitutional role, arguing that the institutional integrity in this age of AI and deepfakes depends on its capacity to stay “accessible and transparent”.

2023 VOICE REFERENDUM AND FRAUDULENT ADS:

During the 2023 Voice to Parliament referendum, misinformation circulated on social media falsely claiming that voting in the referendum was not compulsory. This is a known voter-suppression technique seemingly to discourage voter turnout, although turnout was not measurably affected (Adam, 2024). The mechanism leverages voter's tendency towards inaction, as Max Grömping, an election disinformation political scientist observed, “We see that in more and more elections...message saying, ‘Oh, you know, the election is postponed, it's next week, you don't have to show up.’” (Adam, 2024). Whitty (2025) described the common cyber-tools that amplified mis- and disinformation during this 2023 referendum: automated bots and fake accounts disseminating false narratives that creates an illusion of consensus, deepfakes and artificial intelligence created synthetic media, and “algorithmic exploitation”, defined as the manipulation of platform recommendation systems to prioritise misleading content (Lin 2018; Lindsay 2013; Parks & Duggan 2011).

In November 2023, an AI-generated deepfake of Treasurer Jim Chalmers, Reserve Bank Governor Philip Lowe and other business leaders was created and circulated on Facebook to promote a fraudulent investment platform (Broinowski and Martin 2024, 3). Audio was altered over a real A Current Affair video clip, crafting the figures appearing to endorse and praise the platform, yet in reality an instance of audio being doctored by AI with the intent to mislead. While the subject here was financial fraud rather than electoral persuasion, this case is one of the early documented examples of how convincingly a senior political figure’s image and voice could be easily obtained, synthetically manipulated for paid distribution on a major social platform.

2024:

Deepfakes began to appear in formal political campaigning in 2024. In July 2024, the Liberal National Party (‘LNP’) produced and posted to TikTok a deepfake video of Queensland Premier Stephen Miles dancing, captioned with criticisms of cost-of-living attack lines. The video included an AI disclosure label as disclaimer which the LNP defended on that basis, but Miles was deeply critical of its usage in a political setting, and the Electoral Commission of Queensland noted that current state legislation drew no distinction between human and AI-generated content provided the material was properly authorised (Gould 2024).

A similar use of the technology emerged in September 2024, where Senator David Pocock commissioned and posted deepfake videos of Anthony Albanese and Opposition Leader Peter Dutton expressing support for a ban on gambling advertising which is a position neither leader has taken. This was done to foreground the limited regulation for AI generated videos, for which relevant authorisation was sufficient to not render it illegal (ABC 2024).

In October 2024, in the lead-up to council elections in Victoria, the Victorian Electoral Commission (‘VEC’) through electoral commissioner Sven Bluemmel publicly warned that AI-generated and AI-modified material was expected in campaigns and called for visible AI labelling on all such content. During the election, a series of fake Facebook accounts posted disinformation, advertisements and other posts, endorsing certain candidates and attacking others with the assistance of generative AI. Experts concluded that despite a meaningful number of posts, this was not the work of a well-funded campaign, but that AI has facilitated the ability even for individuals and groups with limited capital to spread misinformation and influence politics, even if only on a local level (Symons and Singh 2024).

2025:

Two ABC News investigations conducted in early 2025 revealed how cheap and convincing the production of political deepfakes has become. ABC News Verify cloned Senator Jacqui Lambie’s voice for approximately 100 AUD using a publicly available platform, which is enough to “fool the software” and many test listeners were unable to identify the recording as synthetic. (ABC 2025) In another parallel experiment, the ABC commissioned this deepfake video of Greens senator David Shoebridge with his permission appearing to advocate for the legalisation of cannabis, which senator Shoebridge described as “uncomfortably persuasive”, which he then emphasised that most viewers encounter such materials in “seven or 10 seconds” on social media where minor artefacts go unnoticed (ABC 2025). Yet, AEC has stated publicly that it does not intend to regulate political deepfakes. Former Deputy Commissioner Jeff Pope told the ABC the AEC “does not have the skills to monitor deepfakes” and that, like policing truth in political advertising, doing so would be outside its remit (ABC 2025).

In May 2025, a Russian-linked network was detected operating under the “Pravda Australia” banner, publishing large volumes of pro-Kremlin material in the lead-up to the 2025 election. This content was directed not at human readers, but AI models, in order for them to consume content and repeat it back to Western users, effectively ‘grooming’ large language models (LLMs) to act in certain ways. By flooding the internet with low-quality, pro-Kremlin articles, they sought to contaminate the AI systems that voters would consult for electoral information (Lavoirpierre and Workman 2025). NewsGuard’s March 2025 audit found that the chatbots operated by the ten largest AI companies repeated narratives traceable to the Pravda network 33.55 per cent of the time (UTS 2025). 

In September 2025, in the context of South Australia’s algal bloom, Liberal candidate Frank Pangallo questioned government officials on a possible link between the bloom and the Adelaide Desalination Plant, providing a list of academic sources to the Budget and Finance Committee with noticeable issues such as incorrect dates and invalid links. After scrutiny from the Labor party, Pangallo admitted that he had used a generative AI application to compile the list (ABC News 2025). This event marks one of the first publicly admitted uses of AI-fabricated citations by an Australian candidate within a parliamentary record. This is alarming in that AI hallucinations not only affect campaign messaging but the evidentiary substrate of political debate, with leading implications for muddling the integrity of Hansard and committee records that voters and journalists treat as authoritative.

In November 2025, a deepfake of premier Roger Cook appeared on YouTube advertisements endorsing another fraudulent investment platform. WA Consumer Protection Commissioner Trish Blake explained that the format of the video as a pop-up made it difficult to assess its validity, and warned that the technology was continually developing and becoming cheaper to weaponise (Shepherd 2025). Despite as with the Chalmers deepfake of November 2023, both usage of deepfakes intended for financial frauds using a credible political figure, yet the operational convenience again indicates that commodity-grade generators are becoming more available for non-political usage and will equally be available for politically motivated deepfakes.

---

REFERENCES

Adam, David. , “MISINFORMATION MIGHT SWAY ELECTIONS - BUT NOT IN THE WAY THAT YOU THINK,” Nature 630, no. 8018 (2024): 807-809, doi:https://doi.org/10.1038/d41586-024-01696-z. https://wwwproxy1.library.unsw.edu.au/login?url=https://www.proquest.com/scholarly-journals/misinformation-might-sway-elections-not-way-that/docview/3074816138/se-2. 

Al Jazeera. 2024. "Did artificial intelligence shape the 2024 US election?" December 25, 2024. 

Al Jazeera. 2024. “How AI Deepfakes Are Influencing Elections Around the World.” Al Jazeera, accessed May 2, 2026. https://aje.io/fpy2i3 

Associated Press. 2023. “Image Claiming to Show Trump Dancing with Underage Girl Is Fake.” AP News. https://apnews.com/ap-fact-check/image-claiming-to-show-trump-dancing-with-underage-girl-is-fake-00000188e8ebdc10ad9bf8eb72850000 

Barnes, Renee, Rory Mulcahy, Lucas Whittaker, and Aimee Riedel. 2024a. “Disinformation and Deepfakes Played a Part in the US Election. Australia Should Expect the Same.” Edited by Sunanda Creagh and Amanda Dunn. The Conversation, November. https://doi.org/10.64628/aa.qaexfv33g. 

Broinowski, Anna, and Fiona R Martin. 2024. “Beyond the Deepfake Problem: Benefits, Risks and Regulation of Generative AI Screen Technologies.” Media International Australia, December. https://doi.org/10.1177/1329878x241288034. 

Bruns, Axel, Daniel Angus, and Timothy Graham. 2021. “Twitter Campaigning Strategies in Australian Federal Elections 2013–2019.” Social Media + Society 7 (4): 205630512110634. https://doi.org/10.1177/20563051211063462. 

Chesney, Bobby, and Citron, Danielle. 2019. "Deep Fakes: A Looming Challenge for Privacy, Democracy, and National Security." California Law Review 107 (6): 1753–1820. 

CNN. 2024. “Fake Joe Biden Robocall.” January 22, 2024. https://www.cnn.com/2024/01/22/politics/fake-joe-biden-robocall 

Dehghan, Ashkan, Kinga Siuta, Agata Skorupka, et al. . 2023. "Detecting Bots in Social-Networks using Node and Structural Embeddings," Journal of Big Data 10 (1): 119. https://doi.org/10.1186/s40537-023-00796-3. https://www.proquest.com/scholarly-journals/detecting-bots-social-networks-using-node/docview/2841403413/se-2. 

Department of Industry, Science and Resources. 2025. National AI Plan. Canberra. https://www.industry.gov.au/publications/national-ai-plan. 

Evans, Jake. 2024. “Senator David Pocock Creates AI Deepfakes of Anthony Albanese and Peter Dutton to Call for Ban ahead of Election - ABC News.” ABC News, September 7, 2024. https://www.abc.net.au/news/2024-09-07/david-pocock-creates-ai-deepfake-of-albanese/104323688. 

Feldman, Dalit K., and Yifat Nahmias. 2025. “From Bots to Ballots: Democratic Integrity in the Era of Digital Manipulation.” Minnesota Journal of Law, Science & Technology 26 (1): 228–[page range if known]. https://doi.org/10.24926/15529541.3905. 

Feldman, Tal, and Pappu, Aneesh. “The Era of AI Persuasion in Elections Is About to Begin.” MIT Technology Review, December 5, 2025. https://www.technologyreview.com/2025/12/05/1128837/the-era-of-ai-persuasion-in-elections-is-about-to-begin/ 

Ferrara, Emilio, et al. 2016. "The rise of social bots." Communications of the ACM 59 (7): 96–104. 

Fu, Hanbi, and Yongqiang Sun. 2024. “Unravelling the Algorithm Manipulation Behavior of Social Media Users: A Configurational Perspective.” ICEB 2024 Proceedings (Zhuhai, China), no. 20. https://aisel.aisnet.org/iceb2024/20. 

Gabor, Eugen, Marian Oancea, and Vladimir Pripp. 2024. "The Impact of Deep Fakes in the Age of Populism and Post-Democracy." Revista De Stiinte Politice (83): 32-46. https://www.proquest.com/scholarly-journals/impact-deep-fakes-age-populism-post-democracy/docview/3110031793/se-2. 

Gould, Courtney. 2024. “Queensland Premier Steven Miles Slams Opposition for Posting AI-generated Attack Ad on TikTok.” ABC News, July 22. https://www.abc.net.au/news/2024-07-22/qld-premier-slams-opposition-for-ai-generated-tiktok/104126936 

Howard, Philip N., Samuel Woolley, and Ryan Calo. 2018. “Algorithms, Bots, and Political Communication in the US 2016 Election: The Challenge of Automated Political Communication for Election Law and Administration.” Journal of Information Technology & Politics 15 (2): 81–93. doi:10.1080/19331681.2018.1448735. 

Islam, M. B. E., et al. 2024. "AI Threats to Politics, Elections, and Democracy." Blockchains 2 (4): 458–481. 

Jesús Pérez Dasilva, Meso Ayerdi Koldobika, and Terese Mendiguren Galdospin. . 2021. "Deepfakes on Twitter: Which Actors Control their Spread?" Media and Communication 9 (1): 301. https://doi.org/10.17645/mac.v9i1.3433. https://www.proquest.com/scholarly-journals/deepfakes-on-twitter-which-actors-control-their/docview/2511143557/se-2. 

Leah MacLennan. 2025. “AI ‘Admin Error’ Puts SA Liberals’ Star Recruit in Spotlight.” Abc.net.au. September 20, 2025. https://www.abc.net.au/news/2025-09-21/ai-liv-golf-algal-bloom-comments-put-frank-pangallo-in-spotlight/105787864. 

Lin, H. 2018. “Developing Responses to Cyber-Enabled Information Warfare and Influence Operations.” Lawfare. 2018. https://www.lawfaremedia.org/article/developing-responses-cyber-enabled-information-warfare-and-influence-operations. 

Lindsay, Jon R. 2013. “Stuxnet and the Limits of Cyber Warfare.” Security Studies 22 (3): 365–404. https://doi.org/10.1080/09636412.2013.816122. 

MacLennan, Leah. 2025. “SA Liberals' Star Candidate in Spotlight after AI Algal Bloom ‘Admin Error’.” ABC News, September 21. https://www.abc.net.au/news/2025-09-21/ai-liv-golf-algal-bloom-comments-put-frank-pangallo-in-spotlight/105787864. 

Manfield, Evelyn, and Matthew Doran. 2025. “Politicians Fear Deepfakes Could Change the Face of This Election Campaign.” ABC News. January 30, 2025. https://www.abc.net.au/news/2025-01-31/politicians-fear-deepfakes-could-impact-election/104876130.  

Martino, Matt. 2025. “We Cloned Senator Jacqui Lambie’s Voice with AI to Show You What a Deepfake Election Could Look Like.” ABC News. February 27, 2025. https://www.abc.net.au/news/2025-02-28/jacqui-lambie-ai-generated-voice-election-and-deepfakes/104986434.  

Newman, Nic, Amy Ross Arguedas, Craig T. Robertson, Rasmus Kleis Nielsen, and Richard Fletcher. 2025. Reuters Institute Digital News Report 2025. Oxford: Reuters Institute for the Study of Journalism. https://doi.org/10.60625/risj-8qqf-jt36. 

Neyazi, Taberez Ahmed, Arif Hussain Nadaf, Khai Ee Tan, and Ralph Schroeder. 2024. “Does Trust in Government Moderate the Perception towards Deepfakes? Comparative Perspectives from Asia on the Risks of AI and Misinformation for Democracy.” Government Information Quarterly 41 (4): 101980. https://doi.org/10.1016/j.giq.2024.101980. 

Paris, Britt, and Joan Donovan. 2022. Deepfakes and Cheapfakes. Data & Society Research Institute. 

Park, Sora, Caroline Fisher, Kieran McGuinness, Jee Y. Lee, Momoko Fujita, Ashleigh Haw, Kerry McCallum, and Gianni Nardi. 2025. Digital News Report: Australia 2025. Canberra: News and Media Research Centre, University of Canberra. https://apo.org.au/sites/default/files/resource-files/2025-06/apo-nid330740_0.pdf. 

Parks, Raymond C., and David P. Duggan. 2011. “Principles of Cyberwarfare.” IEEE Security & Privacy Magazine 9 (5): 30–35. https://doi.org/10.1109/msp.2011.138. 

Parliament of Australia. 2023. “Chapter 2 - Regulating the AI Industry in Australia.” Aph.gov.au. 2023. https://www.aph.gov.au/Parliamentary_Business/Committees/Senate/Adopting_Artificial_Intelligence_AI/AdoptingAI/Report/Chapter_2_-_Regulating_the_AI_Industry_in_Australia. 

Partnership on AI (PAI). 2024. "How the risk of synthetic media affects global election information." Case Study Report. 

Paterson, Thomas, and Lauren Hanley. 2020. ‘Political Warfare in the Digital Age: Cyber Subversion, Information Operations and “Deep Fakes”’. Australian Journal of International Affairs 74 (4): 439–54. Accessed 28 April 2026. https://doi.org/10.1080/10357718.2020.1734772. 

PMC. 2025. "AI-driven disinformation: policy recommendations for democratic resilience." July 31, 2025. 

Politico. 2023. “DeSantis PAC Shares Apparently AI-Generated Fake Images of Trump Hugging Fauci.”  July 17, 2023. https://www.politico.com/news/2023/07/17/desantis-pac-ai-generated-trump-in-ad-00106695 

Ray, Andrew. 2021. ‘Disinformation, Deepfakes and Democracies: The Need for Legislative Reform’. University of New South Wales Law Journal 44 (3). Accessed 28 April 2026. https://doi.org/10.53637/DELS2700. 

Ray, S. 2021. "Deepfakes and the Law." Journal of Cyber Policy 6 (3): 985–987. 

Rebelo, Karen. 2024. “India’s Generative AI Election Pilot Shows Artificial Intelligence  in Campaigns is Here to Stay”. Center for Media Engagement. https://mediaengagement.org/wp-content/uploads/2024/10/Indias-Generative-AI-Election-Pilot-Shows-Artificial-Intelligence-In-Campaigns-Is-Here-To-Stay.pdf  

Rest of World. 2024. “2024 AI Elections Tracker". https://restofworld.org/2024/elections-ai-tracker/   

Reuters Fact Check. 2023. “Video of Hillary Clinton Endorsing Ron DeSantis Is AI Generated.” Reuters, April 2023. https://www.reuters.com/article/fact-check/video-of-hillary-clinton-endorsing-ron-desantis-is-ai-generated-idUSL1N36K1UJ/ 

Robins-Early, Nick. 2023. "Disinformation reimagined: how AI could erode democracy in the 2024 US elections." The Guardian. 

Schick, Nina. 2020. Deepfakes: The Coming Infocalypse. New York: Twelve. 

Schiff, Kaylyn, Schiff, David, and Bueno, Natalia. 2024. "The Liar's Dividend: Can Politicians Claim Misinformation to Evade Accountability?" American Political Science Review 119 (1): 1–20. 

Sheldon, Tony. 2024. “Chapter 2 - Impacts of AI on Democracy.” Aph.gov.au. 2024. https://www.aph.gov.au/Parliamentary_Business/Committees/Senate/Adopting_Artificial_Intelligence_AI/AdoptingAI/Interim_report/Chapter_2_-_Impacts_of_AI_on_democracy. 

Shepherd, Briana. 2025. “Deepfake AI Ad Featuring WA Premier Roger Cook Prompts Latest Scam Warning.” ABC News, November 8. https://www.abc.net.au/news/2025-11-08/wa-premier-deepfake-ai-scam-prompts-warning-amid-cheaper-tech/105987626. 

Somers, Meredith. “Deepfakes, Explained.” MIT Sloan School of Management. July 21, 2020. https://mitsloan.mit.edu/ideas-made-to-matter/deepfakes-explained  

South, 2025. “Libs’ Dodgy Document Scandal Deepens.” Premier of South Australia. September 17, 2025. https://premier.sa.gov.au/media-releases/news-archive/libs-dodgy-document-scandal-deepens. 

Summerfield, Christopher, Lisa P. Argyle, Michiel Bakker, Teddy Collins, Esin Durmus, Tyna Eloundou, Iason Gabriel, et al. 2025. "The Impact of Advanced AI Systems on Democracy." Nature Human Behaviour 9 (12) (12): 2420-2430. doi:https://doi.org/10.1038/s41562-025-02309-z. https://www.proquest.com/scholarly-journals/impact-advanced-ai-systems-on-democracy/docview/3286137422/se-2.  

Symons, Bec, and Tavleen Singh. 2024. “Fake Social Media Accounts, AI Interfere with Local Government Elections." ABC News, November 4. https://www.abc.net.au/news/2024-11-04/ai-fake-social-media-accounts-victorian-local-council-elections/104497330. 

Taylor & Francis. 2025. "Artificial Intelligence in Election Campaigns: Perceptions, Penalties, and Implications." February 19, 2026. 

Taylor, James. 2024. “VEC Calls for Transparency on AI Use at Elections - Geelong Times.” Geelong Times. October 7, 2024. https://timesnewsgroup.com.au/geelongtimes/news/vec-calls-for-transparency-on-ai-use-at-elections/. 

 The Guardian. 2024. “Biden Robocall Indictment in New Hampshire Primary Case.” The Guardian, May 23, 2024. https://www.theguardian.com/us-news/article/2024/may/23/biden-robocall-indicted-primary 

Turner, Brittany. 2025. “Balancing Openness and Integrity: Parliament’s Role in the Age of AI and Deepfakes.” Australasian Parliamentary Review 40 (2): 160–170. https://search.informit.org/doi/10.3316/informit.T2025112100021700876618249 

UTS. 2025. Review of Is Russia Grooming AI Chatbots? UTS. November 21, 2025. https://www.uts.edu.au/news/2025/11/is-russia-grooming-ai-chatbots. 

Wahl-Jorgensen K., Carlson M. (2021). Conjecturing Fearful futures: Journalistic discourses on deepfakes. Journalism Practice, 15(6), 803–820. https://doi.org/10.1080/17512786.2021.1908838 

Weiner, Matthew D. 2024. “Destined to Deceive: The Need to Regulate Deepfakes with a Foreseeable Harm Standard.” Michigan Law Review 122 (4): 771–807. https://doi.org/10.36644/mlr.122.4.destined  

Westerlund, Mika. 2019. "The Emergence of Deepfakes." Technology Innovation Management Review 9 (11). 

Whitty, M. T. , “How the Voice Lost its Voice: Applying the Dual Processing Theory to Explain how Mis/Disinformation can Deceive and Persuade Voting Decision Making,” Journal of Information Warfare 24, no. 2 (2025): 93-108. https://wwwproxy1.library.unsw.edu.au/login?url=https://www.proquest.com/scholarly-journals/how-voice-lost-applying-dmual-processing-theory/docview/3202187078/se-2. 

Woolley, Samuel, and Philip N. Howard. 2016. "Political Communication, Computational Propaganda, and Autonomous Agents." International Journal of Communication 10. 

Yan, H. Y., et al. 2025. "The origin of public concerns over AI supercharging misinformation in the 2024 U.S. presidential election." Harvard Kennedy School Misinformation Review. 

Youngblood, R. 2026. "I Knew You Were Trouble: Deepfakes, Misinformation, and the Threat to Elections." Louisiana Law Review 86. 

Łabuz, M., and C. Nehring. 2024. "On the way to deep fake democracy? Deep fakes in election campaigns in 2023." European Political Science 23: 454–473. 

Łabuz, Mateusz and Grażyna Strnad. 2025. "Political Advertising with the use of Deep Fakes—Can Deliberative Democracy really Benefit from this Strategy?" Politics & Policy (Online) 53 (3) (06). doi:https://doi.org/10.1111/polp.70035. https://www.proquest.com/scholarly-journals/political-advertising-with-use-deep-fakes-can/docview/3229076575/se-2. 

“AI Technical Standard: Statement 4 | Digital.gov.au.” 2025. Digital.gov.au. Australian Government. 2025. https://www.digital.gov.au/policy/ai/AI-technical-standard/ai-technical-standard-statement-4.