Fake It Till You Make It
Political Deepfakes and the Fragility of Authority in Malaysia
By Samantha Khoo, External Contributor
In 2019, a video allegedly depicting Malaysian minister Azmin Ali in a same-sex encounter circulated online. The incident escalated into a national scandal that intersected with factional contestation, moral politics and coalition stability. Digital forensics specialists were called upon to analyse the footage, yet their assessments did little to settle the political debate. Crucially, the political story was not only the content of the footage, but whether it could be trusted at all. Azmin Ali addressed this directly through public denial and media appearances that were themselves recorded and circulated online. In doing so, his rebuttals became part of a digital archive that could later serve as raw material for manipulation or synthetic media.
As synthetic media becomes increasingly sophisticated, political conflict increasingly shifts from disagreement over interpretation to disagreement over reality itself. Even prior to the popularisation of deepfake technology, the Azmin case demonstrated that once digital manipulation becomes socially plausible, disputes over authenticity do not remain technical matters left to forensic experts. Instead, they become political resources that actors mobilise to defend reputations, undermine opponents, or cloud the public sphere with uncertainty, particularly in Malaysia, where sexual scandals have historically shaped moral contestation and political legitimacy.
That uncertainty has intensified with the rise of generative AI, as Malaysia increasingly confronts cases where synthetic media is operationalised as coercion. In September 2025, Malaysian MPs and senators were blackmailed with threats to release AI-generated sexual deepfakes unless US$100,000 was paid, exposing significant legal gaps and concerns about broader institutional trust.
Simultaneously, Malaysia’s political communication ecosystem has become dependent on platform logics, particularly among younger citizens whose political exposure is disproportionately algorithmic, short-form and creator-mediated. Analysts of GE15 described the election as one fought through TikTok strategies and platform culture, while acknowledging that algorithmic reinforcement tends to amplify predispositions rather than invent them organically. In other words, deepfakes are transcending into an environment that is already structurally reliant on mediated authenticity and recognising credibility through cues produced within attention economies.
These developments illustrate an important analytical point that is easy to miss if deepfakes are framed simply as “misinformation”. Reputational harm can arise from circulation, especially in digital environments where screenshots, re-uploads, and archival systems create durable traces that persist beyond verification. Even when synthetic media is later debunked or removed from the original platform, it can still continue circulating through reposts, wayback archives, or private channels that preserve material long after its initial publication. In some cases, controversial content may even be deliberately preserved to keep allegations in circulation. Once suspicion enters the public sphere, corrections or forensic clarification may do little to reverse reputational harm, particularly in polarised environments where uncertainty itself can become politically useful. The asymmetry between the speed of virality and the slower pace of verification means that harm may persist even after authenticity disputes are addressed.
In Malaysia, these effects intersect with an already polarised information environment and a political culture in which reputational narratives, moral framing, and factional contestation shape offline perceptions of legitimacy. Malaysia is consequently entering a condition of synthetic legitimacy, in which political authority remains dependent on visual and audio cues even as the capacity to fabricate them becomes more accessible. Deepfakes should therefore be understood as a crisis in which the credibility of audio-visual evidence itself is in doubt, forcing political systems to operate under structural uncertainty.
What counts as a deepfake, and why does legitimate use matter?
The term deepfake is often used as shorthand for political deception, but refers to a broader family of AI-generated or AI-manipulated media produced through machine learning techniques that synthesise faces, voices, and scenes with increasing realism. Deepfakes may involve manipulated speeches, face-swaps, or synthetic pornography, but the same technical pipeline also supports legitimate applications such as life-like avatars and AI-enhanced creative content. The political risks, therefore, arise from deceptive deployment in contexts where credibility and legitimacy depend on evidentiary media.
Acknowledging the legitimate uses of AI-generated media is, therefore, necessary because synthetic media is not inherently malicious. Generative systems support lawful applications in entertainment, education, and accessibility, making blanket bans difficult to implement given their dual-use nature. As a result, governance challenges stem less from the existence of the technology itself than from how synthetic media is deployed within political and informational environments. Overly broad regulation may therefore produce unintended effects by discouraging legitimate forms of expression, such as satire and parody, while creating uncertainty around acceptable political speech.
This is particularly important in Malaysia, where broader debates surrounding media regulation and online speech governance have already highlighted tensions between addressing harmful content and safeguarding legitimate political expression and public discourse. In such an environment, vague or expansive deepfake regulation could create significant enforcement discretion and uncertainty for journalists, activists and political actors. Existing laws on communications, fraud, and obscenity can already be applied to harmful uses of synthetic media, even as newer initiatives such as the Online Safety Act 2025 remain in transition. The broader challenge is not simply regulating the technology itself, but responding to the political and informational conditions that shape how synthetic media is used, interpreted, and contested.
From social media election to algorithmic contestation
Malaysia’s experience with platform politics did not begin with deepfakes, and this historical trajectory matters because it shows that political legitimacy had already begun migrating toward visibility-based signals long before synthetic media entered the picture. Malaysia is among the most digitally connected societies in Southeast Asia, with internet penetration reaching 97.4 percent and social media usage at 83.1 percent of the population. Under such conditions, much of the country’s political conversation unfolds within digitally networked environments rather than through discrete campaign events or traditional media cycles.
This shift became particularly visible during the 2013 general election. GE13 has been widely described as Malaysia’s first “social media election”, an expression that captured the growing strategic role of online communication even at a time when Facebook, not TikTok, dominated the digital landscape. Since then, social media has reshaped Malaysia’s election landscape by enabling alternative narratives, facilitating networked mobilisation, and weakening the monopoly power of traditional media, even as digital spaces also created new vulnerabilities to polarisation and coordinated manipulation. Over time, platforms became more than mere communication tools, as political actors began adapting their strategies and legitimacy performances to the forms of visibility and engagement that digital platforms reward.
By GE15 in 2022, TikTok was described as a central political arena for youth outreach, with parties deploying short-form video strategies to generate relatability and cultural fluency among younger voters who were newly electorally consequential under UNDI18. This reflected recognition that contemporary campaigning had become intertwined with platform culture. The objective was to appear authentic within the aesthetic norms of the platform, often at the expense of substantive policy communication.
Yet the political influence of TikTok should not be understood as a mechanism that converts online popularity into entirely new electoral constituencies. Rather, social media platforms function primarily as amplification systems that make pre-existing political sentiment more salient within public discourse. Algorithmic recommendation systems reinforce ideological predispositions by repeatedly exposing users to content that aligns with their existing interests and engagement patterns. Repeated exposure also normalises particular narratives and interpretations of political reality. The political significance therefore, lies in selective visibility, or the gradual process through which particular framings become recognisable as credible or authoritative through sustained presence within users’ feeds.
Youth politics in the deepfake era
Deepfake risk is often framed through a deficit model that assumes young voters are inherently gullible. A more accurate framing is that vulnerability is structural. Youth political exposure in Malaysia increasingly unfolds within algorithmically curated feeds. As a result, political information circulates less through structured debate and more through short-form videos, memes, and creator-driven interpretations that blur the boundaries between entertainment and political communication. In this context, deepfakes do not need to convert voters ideologically; they can instead sharpen affect, intensify cynicism, or reinforce distrust by making the information feel unreliable.
When cues used to establish authenticity can be cheaply manufactured, trust declines. This particularly affects younger voters who rely heavily on platforms to distinguish between authentic and synthetic media, requiring forms of scrutiny that political participation was not designed to accommodate.
Over time, this instability can produce two corrosive outcomes. Faced with persistent uncertainty, some users may adopt generalised scepticism in which all political information is treated as potentially fabricated. Others may retreat into partisan information environments where credibility is determined primarily by group affiliation rather than evidence. In both cases, the result is a weakening of the shared informational foundations upon which democratic deliberation depends.
Hybrid media systems, political influencers, and the re-signalling of authority
These visibility dynamics are reinforced by hybrid media systems in which communication circulates across interconnected networks of journalists, political actors, influencers, and digital creators. Within such environments, authority is no longer conveyed solely through institutional position or traditional media visibility but is mediated by actors who interpret political developments for audiences through platform-native formats. As a result, the boundaries between journalism, commentary, activism, and campaigning have become blurred, producing environments in which political meaning is constructed through interactions between institutional actors and digitally mediated intermediaries.
Research on hybrid media environments shows how influencers operate within these networks and reshape who gets to frame political reality. Within the Malaysian context, policy research has similarly examined the growing prominence of political influencers as intermediaries for audiences that consume political information through creators and micro-media rather than traditional news institutions. Rather than simply relaying political messages, these actors contextualise and personalise developments in ways that can make complex issues more accessible.
The implication is that political legitimacy within platform environments depends on signals of authenticity, such as perceived sincerity. It is within this environment that deepfakes become destabilising. By replicating the cues used to signal authenticity, synthetic media weakens the public’s ability to treat those cues as reliable evidence, eroding the foundations that sustain platform-mediated legitimacy.
When evidence becomes contestable
Malaysia’s emerging condition as one of “synthetic legitimacy” recognises that the deepfake problem cannot be reduced to misinformation or cybercrime alone. What is destabilised is the system through which political authority is recognised. In platform politics, legitimacy increasingly depends on visibility and recognisable forms of authenticity, as political actors adapt to platform-native formats and short-form media. While this can make politics more accessible, it also shifts legitimacy toward performative cues that are more vulnerable to manipulation.
In such an environment, synthetic media attacks distort information and undermine the credibility through which political authority is recognised. Authority can be attacked through fabricated scandal, but it can also be defended through plausible denial, as actors dismiss authentic evidence as manipulated. Known as the “liar’s dividend”, the mere possibility of manipulation allows genuine evidence to be contested or dismissed as fake, leaving audiences less able to treat any media source as credible proof. Institutions that normally adjudicate authenticity, such as courts, law enforcement, or digital forensics, often operate too slowly or remain politically contested to resolve disputes before public narratives have already solidified.
The implications extend beyond reputational harm. Legitimacy in democratic systems depends not only on electoral outcomes but also on public confidence that political speech can be reliably attributed to identifiable actors, that scandals are grounded in verifiable evidence, not fabricated content, and that citizens share a common informational foundation. This means that there would be a shared set of facts, sources, and standards of verification that allow political disagreement to take place within a recognised reality rather than collapse into disputes over whether events occurred at all. When audio and video recordings, which have traditionally functioned as key forms of political evidence, can be convincingly manipulated through synthetic media, establishing that attribution becomes significantly more difficult. Under such conditions, citizens may no longer be able to rely on familiar cues to determine whether a statement or recording is genuine, weakening the shared basis on which political claims are assessed. As these foundations erode, the credibility of the political system itself becomes more fragile.
Governing when authenticity is programmable
If the challenge posed by deepfakes is structural, responses must address the informational conditions under which political legitimacy is constructed rather than focusing solely on individual instances of harm. This requires a combination of legal clarity, platform responsibility, and public capacity.
Governance cannot rely exclusively on reactive response. The central challenge is ensuring that manipulated media can be identified, contextualised and addressed quickly enough to prevent the erosion of public trust. Thus, platforms play a particularly important role because the speed of virality often far exceeds the pace of institutional response. Mechanisms such as rapid takedown pathways, transparency requirements for political content, and clear labelling of synthetic media may become essential tools for limiting harm while preserving legitimate creative uses of generative technologies.
Equally important is developing citizen and institutional capacity to verify political information through trusted sources and transparent authentication mechanisms. This requires moving beyond general media literacy toward practical toolkits that help users assess credibility in real time. This may include clearer provenance signals for digital media, such as watermarking or content origin indicators, as well as accessible verification channels through official institutions and independent fact-checking bodies.
Strengthening the evidence-based environment requires reinforcing the visibility and reliability of authoritative sources. This may involve standardising verification markers for official communications, improving the discoverability of primary sources, and ensuring that corrections or clarifications are disseminated with comparable reach to the original content. Without such measures, the issue is not simply that false content circulates, but that the line between credible and non-credible claims becomes increasingly blurred.
Malaysia’s experience illustrates a challenge that will likely confront many political systems across Southeast Asia. As generative AI tools become more accessible and synthetic media becomes more convincing, the problem will not simply be identifying individual deepfakes but maintaining the informational foundations upon which democratic legitimacy depends. When authenticity becomes programmable, the resilience of political institutions may ultimately depend on whether societies can build systems of trust that are harder to fake. In an era of synthetic media, political authority may increasingly hinge on who can most convincingly “fake it until they make it”.