The Infinite Slop Machine

The Infinite Slop Machine

Executive Summary

With the rise of AI and social media, democracies across the world are facing ever more sophisticated influence operations. While their operators and objectives might differ, the methods they use and vulnerabilities they exploit can often be the same.

In this report, democracy defenders from the Asia Pacific and Europe joined forces to understand the techniques being used and vulnerabilities being exploited by two nearly identical operations.

While the target of the operation might differ, it’s in the best interest of maintaining a high-integrity information environment to identify this style of low-quality content and patch the vulnerability that enables it. Democracies worldwide face a deluge of low-quality political content, and cross-national and even continental collaboration is essential to effectively tackle this problem.

This report documents a coordinated influence operation on YouTube consisting of at least two closely related networks of accounts publishing large volumes of AI-generated geopolitical content. In total, the operation involved more than 29 accounts and over 7,300 analysed data points from videos published between March and December 2025.
The network relies heavily on automated video production using the InVideo platform. Videos typically combine AI-generated narration, synthetic thumbnails, stock footage, and templated video formats. Channels publish at a high frequency, generally producing approximately two videos per day. The case is similar to the influence operation Shadow Play in that it rapidly produces content on YouTube and pushes narratives around specific geopolitical topics.

Analysis identified two main clusters. The first cluster appears to target Canadian audiences, often publishing content critical of U.S. tariff policies toward Canada. The second cluster consists of channels using “Hub” or “Updates” naming patterns that focus on a broader range of geopolitical topics.

Across both clusters, recurring narratives include claims about the growing strength of BRICS, portrayals of the European Union as weakening, and geopolitical coverage related to the Philippines and regional tensions in Southeast Asia. In some cases, the channels circulated false claims, including videos alleging naval clashes between the Philippines and Malaysia that were publicly denied by both navies. Content from these channels is frequently redistributed beyond YouTube, particularly on Facebook and external blogs, extending the reach of the narratives across platforms.

As of April 2026, four of the accounts are no longer accessible since this investigation began, and one channel has changed its name. Overall, the activity documented in this report exhibits multiple indicators of coordinated inauthentic behaviour. While the precise motivation behind the network remains unclear, the scale of automation and the synchronised publishing patterns suggest a deliberate operation that warrants further investigation.

Not addressing AI-enabled content laundering likely puts YouTube in breach of its DSA obligations. As a Very Large Online Platform, it must identify systemic risks (Article 34) and mitigate them effectively (Article 35).

This qualifies as a systemic risk because it manipulates recommendation algorithms and enables coordinated inauthentic behaviour at scale. Since the same techniques were already documented in the Chinese "Shadow Play" influence operation, YouTube should have anticipated them.

Crucially, removing accounts is not enough: AI enables new channels to be created almost instantly, so YouTube needs broader, systemic countermeasures such as detecting semantically equivalent content and identifying coordinated posting patterns across accounts.

Key Findings:
- At least 29 coordinated YouTube accounts were identified.
- The network published over 7,300 videos between March and December 2025.
- Channels used automated AI video generation tools (InVideo, AI voiceover, templated visuals).
- Accounts uploaded roughly 2 videos per day per channel.
- Content was frequently amplified on Facebook and external blogs.
- Several videos promoted verifiably false geopolitical claims.