Disinformation & misinformation, explained.
Despite widespread use of the term, the exact definition of AI slop continues to evolve (Getty Images)
The line between commercial AI slop and state propaganda is blurring, and platforms are struggling to keep up.
Subscribe to The Informer for monthly expert analysis, and to Events for advance notice of visiting world leaders and distinguished guests.
You may unsubscribe from Lowy Institute newsletters at any time. For information on our privacy practices and how to unsubscribe, see our Privacy Policy.
Hostile influence peddling, slopaganda style Hostile influence peddling, slopaganda style
Recent news (Opens in new window) of Australian researchers uncovering a foreign network of hundreds of Facebook pages with AI deepfakes impersonating Australian politicians was a surprise to many. It was revealed the campaign was linked to an influencer business in Sri Lanka, with 10% of the network’s pages monetised by Facebook. Similar influence campaigns were also prevalent (Opens in new window) in the ongoing Iran war, with AI-generated slop being used to produce misleading content on the conflict.
Influence campaigns utilising AI slop have also been observed in other parts of Asia. In Singapore, an AI-driven disinformation campaign was uncovered, involving hundreds of YouTube videos disseminating false and misleading information about Singapore’s foreign policy, economy and leadership. Journalists who uncovered the campaign (Opens in new window) found that multiple YouTube accounts involved in the operation appeared to be centrally coordinated, deemed “a hallmark of influence operations to manipulate public opinion”.
But how is AI slop defined? Experts typically refer to AI slop (Opens in new window) as high-volume synthetic content (text, images, video, audio) generated by AI and optimised for engagement. Some examples of slop content include repetitive memes, auto written articles or uncanny visuals posted in bulk. Despite widespread use of the term, the exact definition of AI slop continues to evolve.
The same infrastructure used by commercial actors for profit and financial gain, however, can be repurposed by hostile actors.
AI slop containing propagandistic elements is known as “slopaganda” – a portmanteau of the words “slop” and “propaganda” – which is the use of AI-generated slop that serves propagandistic purposes (Opens in new window).
Common characteristics of AI slop include the high volume of content generated by AI, with repetitive content to shape and influence the beliefs or views of individuals. Most AI slop is produced for commercial and entertainment reasons. Generative AI makes it possible for content creators to produce hundreds of videos or images at minimal cost, while platform recommendation and monetisation systems reward content that captures attention. The resulting material can be repetitive and inaccurate without being deliberately deceptive.
AI slop is one manifestation of a wider trend towards the industrialisation of synthetic content. The same infrastructure used by commercial actors for profit and financial gain, however, can be repurposed by hostile actors.
AI slop can become a force multiplier for hostile information campaigns such as coordinated influence operations and election misinformation – campaigns that are, by definition, purposeful and directional in the pursuit of a strategic objective.
State and non-state actors may disguise and weaponise AI slop to make it indistinguishable from commercial slop infrastructure, and disguise or conceal the propagandistic narratives in commercial AI slop.
In other words, AI slop does not inherently signify the presence of an influence operation. Rather, AI slop, especially AI slopaganda, can be a tool within an influence operation.
Markus Spiske/Unsplash
Prior research by Graphika in 2025 revealed (Opens in new window) that online propaganda campaigns are created with AI for core functions such as content creation and creating inauthentic influencer personas on social media. Examples include fake social media profiles, featuring inauthentic synthetic news reporters in videos, and fake news websites with content generated by AI.
A hostile actor seeking to influence countries in Southeast Asia does not necessarily need to fabricate an entirely new controversy. It could take existing social, political or economic fault lines and generate enormous volumes of content such as AI slop along these narratives.
For instance, in Singapore, journalists uncovered AI-generated female personas on TikTok accounts with reused voices and recycled scripts. Disinformation narratives in the videos framed Singapore as being economically reliant on China. These accounts blended genuine news with false and misleading claims (Opens in new window) on Singapore’s economy. Investigations revealed that there were signs of coordinated behaviour behind the accounts, such as publishing multiple videos within a short period of time. Similar narratives were also found on YouTube. Beyond Singapore, there were also videos that targeted audiences in Malaysia.
The goal of such campaigns utilising AI slop may not entirely be to make a target audience believe a particular falsehood. Rather, the objective may instead be to shape the broader information environment in which citizens form opinions. Seeing the same narratives on multiple videos can create familiarity over time with repeated exposure. The result could contribute to greater uncertainty and distrust, where citizens are flooded with content whose origins they cannot verify. This erodes the broader information environment online.
Tech and social media companies are sharpening their stance on AI slop. Platforms such as YouTube and LinkedIn are removing low-quality spam and updating company guidelines (Opens in new window) to limit the spread of AI slop on their platforms. Other companies such as Meta and TikTok are applying automated AI information labels (Opens in new window) to tag synthetic media.
More companies are incorporating (Opens in new window) SynthID standard of detection to enable individuals to check if images are generated by AI. However, enforcement across the platforms and products remains varied and inconsistent, while algorithmic feeds continue to reward AI-generated posts with high engagement rates.
A layered approach to resilience remains the utmost priority. At the government and institutional level, monitoring of recurring narratives and coordinated networks can be further enhanced. Continued media and information literacy, especially on the concerns and risks of AI-generated content, should be a top priority for governments, educators and civil societies alike. For countries, the challenge goes beyond detecting AI slop, but to enhance a society’s resilience to the narratives and influence operations that AI-generated content can amplify.
About the author
Dymples Leong
Dymples Leong is an Associate Research Fellow with the Centre of Excellence for National Security (CENS) at the S. Rajaratnam School of International Studies (RSIS), Nanyang Technological University, Singapore.
The most-pressing world events explained by Lowy Institute experts and global contributors, in your inbox, every Thursday.
You may unsubscribe from The Interpreter at any time. For information on our privacy practices and how to unsubscribe, see our Privacy Policy.
