Maintaining trust in an era of AI Slop

On 17 September, the ABC News Verify team published an article exposing a political youtuber, Oceania Unfiltered, who published videos “almost exclusively”¹ about Australia and which frequently promoted One Nation. The exposé showed that the channel’s presenter, without declaring it, used generative AI to change his accent and lingo to that of an Aussie. The AI also changed his lips and face to match.

Later the same day, Oceania Unfiltered published a response to the investigation saying that the Youtuber just wanted to make the videos “more pleasant”² for Australian viewers. Whatever the motivation was, it is clear the ABC investigation has had an impact. The channel has now pivoted. All videos from before the 17 September have been deleted or hidden, and the focus of the channel is now predominately on Trump and the United States.
 
Low-quality YouTube videos that are substantially or fully created with generative AI, aka AI-slop, are ubiquitous with one study³ finding that they make up more than a fifth of the videos on YouTube. YouTube is aware that having a huge amount of AI slop on the platform may present a problem. Recently, YouTube CEO Neal Mohen promised that the platform would crack down⁴ on “AI slop.” And YouTube's VP of Trust & Safety, Matt Halprin also made clear⁵ that channels could be demonetised or barred from the YouTube Partner Program⁶ if they publish videos of AI personas that discuss serious topics such as finance, law or healthcare. 

YouTube’s AI disclosure⁷ applies to everyone uploading to the platform and requires content creators to disclose if their video “is fully or partially altered using AI tool” and whether it makes a real person appear to say or do something they didn’t do or alters footage of a real event or place or generates a realistic scene that didn’t actually occur or creates music that’s the main focus of the video. Content creators self declaring any of these will have either a small AI badge or a label affixed to their videos. But judging by the comment sections of many AI generated videos, many people may be missing them. 

So how does AI impact trust on YouTube? On trust and AI, Dr Timothy Koskie⁸ from Centre for AI Trust and Governance at the University of Sydney says “We rely on ‘trusting’ intermediators all the time because we just can't experience it all ourselves. AI-inflected videos can extend existing trends like beauty filters or just plain clever editing, but what we ultimately rely on is less the product than our belief, confidence, or trust in the source.” 

However, he adds that “If AI tools are used to make you think you are seeing your trusted source when it is a fraudulent recreation, that's where we could start to see some real damage.” For this reason, labels on AI generated videos are important and equally as important is that viewers are able to easily see these labels.
 
For serious youtubers and platform journalism creators, there are many challenges. AI sloptuber channels as they are known, have quickly become major competitors in the attention economy. For many creators that have high ethical and editorial standards,⁹ a video is often the result of days, weeks, and sometimes months of research and factchecking. For GenAI channels that don’t bother with such rigor, a previously unimaginable rate of output is possible.  Many can produce several long-form videos every day. 

Additionally, more and more serious youtubers are making efforts to be as transparent as possible regarding their use of AI and many go well beyond the minimum requirements of YouTube or indeed, any regulatory body. As I detailed in a past CMT newsletter,¹⁰ popular youtuber Hank Green’s AI policy was a reaction to his audience’s discovering he used AI use for script writing. But others are actually more proactive. For example, popular science-focused youtubers Simon and Dr Becky¹¹ have recently declared AI editorial policies. A common, explicitly stated part of these AI policies is that AI tools are not used for creating video, images, music or scripts. This doesn’t mean more serious You Tubers aren’t using AI to generate captions, vertical clips from a horizontal video and to analyse expected performance of scripts, titles and thumbnails. But considered policy and transparent application is what distinguishes these creators from the irresponsible ones.

References 

1. https://www.abc.net.au/news/2026-09-17/ai-generated-slop-economy-brazilian-deepfake-alternates/107158050 
2. https://youtu.be/QMYPnyD0lyg 
3. https://www.theguardian.com/technology/2025/dec/27/more-than-20-of-videos-shown-to-new-youtube-users-are-ai-slop-study-finds 
4. https://www.forbes.com/sites/danidiplacido/2026/01/23/youtube-plans-to-fight-against-ai-slop-by-using-ai/ 
5. https://youtu.be/14Vm0CiyUVE 
6. https://www.youtube.com/creators/earn/youtube-partner-program/ 
7. https://profiles.sydney.edu.au/t.koskie 
8. https://www.uts.edu.au/news/2026/08/the-serious-standards-of-serious-youtubers 
9. https://www.uts.edu.au/news/2026/08/the-serious-standards-of-serious-youtubers 
10. https://youtube.com/shorts/9Ftnvqm0MXc 
11. https://www.youtube.com/post/UgkxCIjWdKStTogt3Ir_wx8Vu6ZaqVZgx4Ub