The Evolution of Influence: How AI Slop Is Transforming the Battlefield of Public Opinion in Southeast Asia
The digital landscape of Southeast Asia, a region characterized by its vibrant social media usage and deep political complexities, is facing an unprecedented and insidious threat that is reshaping the very nature of modern warfare. According to prior research conducted by the network analysis firm Graphika in 2025, the days of vast, human-operated troll farms painstakingly crafting fake personas are fading. Instead, state-sponsored and non-state hostile actors are increasingly harnessing the raw power of generative artificial intelligence to conduct online propaganda campaigns at a scale and speed previously unimaginable. This new modus operandi relies on AI for core functions, including automated content creation and the generation of entire inauthentic personas designed to infiltrate social media ecosystems. The tools of this trade are becoming increasingly visible to researchers and intelligence agencies: fabricated social media profiles that look astonishingly real, synthetic news reporters with lifelike yet entirely fake faces delivering scripted diatribes via video, and a sprawling network of bogus news websites spewing articles churned out by large language models. This evolution signals a critical paradigm shift—from a battle for the truth to a war waged through the sheer propagation of algorithmically generated noise, where the volume of nonsense substantially outweighs any factual rebuttal.
However, the most dangerous aspect of this new threat matrix is that hostile actors making the strategic calculation to target countries in Southeast Asia do not always need to spend time or resources fabricating entirely new controversies or inventing wild conspiracies from scratch. Instead, they have adopted a highly efficient and insidious strategy of “weaponizing the existing.” They exploit the already volatile social, political, economic, and religious fault lines that cut deep into the fabric of these societies. By identifying pre-existing grievances—such as economic disparities, ethnic tensions, or contentious geopolitical dynamics—they can deploy generative AI to produce an enormous volume of content, known in the digital vernacular as “AI slop,” that pushes these narratives toward a boiling point. A quintessential example of this strategy was uncovered by journalists in the city-state of Singapore. Investigative reporting revealed a network of TikTok accounts featuring seductive, AI-generated female personas that had captured the attention of the platform’s users. Upon closer inspection, it became clear that these digital “influencers” were not who they claimed to be; they were constructed using reused voices and recycled scripts, the hallmarks of a coordinated automated operation. The propaganda narratives pushed in these videos were carefully crafted to frame Singapore as economically dependent on China, weaving a web of misleading geopolitics that leveraged a genuinely sensitive area of public debate to settle a disinformation objective.
These operations were not limited to simple social media posts; they represented a sophisticated blending of fact with fiction designed to manufacture credibility for the false claims. The AI-generated personas—often presenting as attractive, relatable women—shared videos that mixed genuine news clips, covering real economic data or political decisions, with false, misleading, or heavily distorted claims regarding Singapore’s sovereign wealth funds and its economic future within the broader Sino-American competition. This merging of truth and fiction is a critical element of modern propaganda, as it makes it significantly harder for the average citizen to identify the falsehoods without extensive fact-checking. Furthermore, investigators noted the presence of clear signs of coordinated behavior behind the accounts—a digital firing squad that published multiple videos in rapid succession to saturate the feed and game the algorithmic recommendations. The disinformation campaign did not stop at Singapore’s borders. Similar video content, employing the same techniques and echoing the same anti-Singapore or China-centric themes, was found targeting the broader Malaysian audience, indicating a regional influence operation designed to foster regional instability and sour the public’s perception of national resilience and independence.
To understand this strategy, one must reframe how we perceive the ultimate goal of such influence campaigns. The modern purveyor of AI slop may not necessarily achieve the objective of convincing a target audience to believe a specific, outrageous falsehood—such as the idea that a country is about to default on its debts or that a foreign power has taken control of the local police force. The true objective is far more pernicious and sophisticated: to shape the broader information environment in which citizens form their opinions, thereby eroding the very foundation of social trust. This is a psychological operation targeting the cognitive infrastructure of the state. When citizens see the same divisive narratives circulating across multiple social media platforms—pushed by A.I. avatars, bots, and spoofed news sites—those narratives begin to feel familiar. This is the “illusion of truth” effect, amplified by automation. Over time, this persistent drip of content contributes to a massive sense of uncertainty and confusion. Citizens are flooded with information whose origins and veracity they cannot verify, leading to a state of cognitive dissonance and, eventually, deepest distrust. This erosion of the information ecosystem means that even when legitimate news outlets report the truth, a significant portion of the population may dismiss it as just another piece of propaganda, equivalent to the AI slop they see online. When this occurs, the result is a society where evidence-based reasoning is devalued, and national cohesion is broken down, making the region more susceptible to internal division and external manipulation.
In response to this escalating crisis, tech and social media companies—often regarded as the disinformation battlefields of the 21st century—are finally being forced to sharpen their stance against the rising tide of AI slop. Platforms such as YouTube and LinkedIn have faced intense scrutiny for hosting low-quality, spam-like content and have recently announced sweeping policy changes and visible removals of millions of such videos and posts. These platforms are updating their company guidelines to explicitly prohibit the monetization and spread of inauthentic, automated content that could be used to mislead. Simultaneously, industry giants like Meta (the parent company of Facebook and Instagram) and TikTok are implementing automated AI information labels, tagging synthetic media so that users can see when an image, video, or audio file has been manipulated by generative intelligence. Furthermore, recognizing the lesson that a rising tide of fake content requires a unified defense, more companies are incorporating the SynthID standard of detection—a watermarking technology developed originally by DeepMind—into their products. This enables users and researchers to verify whether an image or text has been generated or altered by AI, adding a critical layer of forensic capability to the fight against automated influence. This regulatory shift toward transparency is a welcome sign, signaling that Silicon Valley is conceding that the problem is real, pervasive, and uniquely amplified by the advance of large language models.
Yet, despite these proactive measures, the enforcement of these new rules across platforms and diverse products remains strikingly varied and inconsistent. The current moderation systems are largely reactive, playing a game of whack-a-mole against an adversary that generates content with the click of a button. While SynthID and other detection tools exist, they are not mandatory for all AI developers, and well-funded hostile actors can easily use open-source models without similar safeguards, meaning the labeled content represents only a fraction of what is out there. Perhaps the most glaring vulnerability lies in the feeds themselves. Algorithmic recommendation systems, which prioritize engagement, reach, and watch time, continue to reward AI-generated posts that evoke strong emotional reactions—outrage, fear, or anger—with extraordinarily high engagement rates. This creates an economic and behavioural incentive for the spread of AI slop, ensuring that even as platforms remove flagged content, the algorithms are simultaneously admitting new, mutated versions into the feed. The malevolent genius of the modern disinformation operation lies in its decentralized nature; it is cheap to run, easy to generate, and can rapidly morph to evade detection within hours of a new policy being put in place. Consequently, the battle against AI slop cannot be won solely by updating terms of service or deploying watermarking algorithms; it demands a permanent, relentless commitment to patching the holes, though the attackers retain the advantage of initiative and scale.
Given that the emergence of this AI-powered uncertainty represents a threat to national security, a layered approach to resilience remains the utmost priority for governments, institutions, and individuals. The challenge is no longer simply about identifying and removing a few fake accounts; it is about building a society that is inherently hostile to the influence of disinformation. At the government and institutional level, monitoring of recurring narratives, cross-platform orchestration, and coordinated networks can be further enhanced through public-private partnerships. National security agencies must collaborate with academic institutions and intelligence-sharing coalitions to map the origin of these AI networks, even as they become increasingly sophisticated. However, the most critical bulwark against this insidious threat lies in the minds of the citizenry itself. Continued, aggressive, and frequently updated media and information literacy must be normalized and woven into the educational curriculum in schools, universities, and vocational training. It is not enough to teach citizens how to spot a deepfake or a bot; the population must be equipped with the critical thinking skills to understand the intent behind algorithmic suggestions, the motivations of state actors, and the importance of verifying sources across a fragmented information ecosystem. For Southeast Asian countries, the challenge is existential and goes far beyond the technical detection of AI slop. It is fundamentally about enhancing a society’s resilience to the narratives, divisions, and influence operations that AI-generated content can amplify, ensuring that democratic institutions, public trust, and social cohesion survive the onslaught of a duplicative, plastic, and deep-fueled future.