Artificial intelligence significantly reduces the cost and time required to produce video. Across major digital platforms, that shift coincides with a notable increase in low-quality, AI-generated content designed to perform well within algorithmic recommendation systems.

New research from Kapwing, an online video editing platform, documents the global rise of what many creators and technologists refer to as AI slop. Rather than defining this content by topic or format, the report focuses on how it is produced and distributed. AI slop is created quickly, published at high volume, and optimized to trigger engagement metrics that recommendation systems use to rank and surface content.

Kapwing’s analysis places AI slop within the mechanics of platform distribution, helping clarify what drives visibility. Across YouTube, TikTok, Instagram, and Facebook, algorithmic systems prioritize scalable engagement signals rather than editorial attributes. Understanding those incentives is essential for publishers to assess where they remain exposed to algorithmic churn and where alternative approaches to audience growth can still exert leverage.

Share of AI slop in recommended video feeds

Kapwing’s analysis of YouTube recommendations finds that a substantial portion of videos shown to new users qualifies as low-quality AI-generated content. In recommendation feeds generated for accounts with limited viewing history, AI slop appears repeatedly rather than as isolated suggestions.

The research identifies entire channels devoted exclusively to producing AI-generated videos at scale. These channels rely on automated workflows to generate visuals, narration, and scripts, often publishing multiple videos per day. Some achieve significant reach and generate revenue through standard platform advertising programs.

Kapwing’s findings suggest that this visibility reflects how recommendation systems function when personalization data is limited. For new accounts, platforms rely more heavily on generalized engagement signals to populate feeds. AI slop frequently meets these criteria, increasing the likelihood that it will be surfaced early and often.

The content itself commonly features recycled visuals, synthetic voiceovers, loosely assembled scripts, and broad or ambiguous topics. Kapwing emphasizes that AI slop is not defined by the use of artificial intelligence alone. Instead, it reflects a production approach designed to maximize engagement metrics such as watch time, posting frequency, and volume.

The earmarks of AI slop

The report distinguishes AI slop from other uses of AI in media production. Many publishers and creators use AI tools to support editing, translation, accessibility, or workflow efficiency. AI slop, by contrast, involves minimal editorial intervention and relies on automation to scale output rapidly.

The defining characteristic of slop is not automation itself, but the absence of editorial oversight. Content decisions are driven primarily by performance data rather than by subject expertise, reporting, or narrative intent. This distinction allows Kapwing to identify slop based on observable production and publishing behaviors.

Signals prioritized by recommendation systems

This research highlights the role of engagement signals in determining content distribution. Across algorithmically curated platforms, recommendation systems rely on metrics that are easy to measure and compare at scale. These include watch time, retention, posting frequency, and consistency.

Editorial attributes such as accuracy, sourcing, originality, and narrative structure do not directly factor into these systems. Their exclusion reflects platform design choices about which signals are incorporated into ranking models.

AI slop is produced to generate high watch time, frequent posting, and consistent engagement. These are the same signals recommendation systems to rank and surface content. High posting frequency increases the likelihood of repeated exposure, particularly in feed-based environments.

Global distribution patterns

Kapwing’s findings show that AI slop is not limited to a single market or language. The report identifies similar patterns across regions and content categories. Channels producing AI slop appear in multiple countries and serve audiences in diverse linguistic contexts.

This distribution reflects shared platform systems rather than localized editorial practices. Where algorithmic recommendation governs visibility, similar outcomes emerge regardless of geography.

From this perspective, AI slop is not an anomaly or a fringe category. It represents a production strategy that performs well within existing algorithmic distribution systems. The research clarifies the distribution dynamics that allow AI slop to scale. This content’s prevalence across platforms, regions, and formats reflects shared incentive structures rather than changes in audience demand.

For media executives, the spread of AI slop underscores a hard truth about today’s distribution economics: platform visibility is driven by scalable engagement signals, not by editorial judgment or quality. When scale and engagement signals drive distribution, automated content gains structural advantages. Publishers must continue to reduce reliance on feed-driven discovery by strengthening direct, first-party audience relationships and prioritizing formats that build habitual, intentional consumption.