The AI Slop Economy Runs on Unpaid Verification

The AI Slop Economy Runs on Unpaid Verification

A growing concern is emerging around artificial intelligence and trust. As AI floods the internet with cheap, plausible content, genuine human-created material becomes scarce, leading to the belief that credibility is the new moat. However, further research reveals a different reality. While the flood of AI-generated content is real, with half of new articles now primarily AI-generated, nobody is paying more for the alternative.

Research shows that more than half of daily uploads on music platforms like Deezer are fully AI-generated, yet these tracks earn only one to three percent of streams, indicating a twenty-to-one decoupling between production and consumption costs. Similarly, in the book industry, while AI text accounts for 14% of Google's ranking set and 11.3% of revenue, titles with substantial AI text make up only 20% of the catalog and 11.3% of revenue. Among the top five percent of bestsellers, 73% contain no AI text at all.

The crumbling economics of human-written content is evident in the data. Revenue per title fell in seven of eight genre clusters for books without AI text, while revenue for AI-generated titles rose 19.2 times. However, this does not create a premium for the authentic; instead, it compresses the economics for everyone. A notable example is Shutterstock, whose business relies on verified, rights-cleared, provenance-documented human-made content.

In Q2 2026, the company's revenue fell by 17%, and its licensing line experienced a 28% decline, leading to a net loss of $155.9 million against $29.4 million in net income a year earlier.

The cost of producing a plausible artifact has dropped to zero, while the cost of verifying one remains unchanged. Moreover, there is no premium for verification, as it becomes an expense incurred by those closest to the process. This shift is evident in the global vulnerability pipeline, where security researchers found that a single GitHub account filed 55 fraudulent vulnerability reports, all of which passed through the system undetected.

Institutions built on the assumption that verification adds value are now facing the consequences. GitHub shut down a seven-year-old bug bounty due to its collapsing confirmation rate, while arXiv began requiring prior peer review for survey papers. Wikipedia added a speedy-deletion criterion for AI-generated drafts, and the National Vulnerability Database moved its backlog of pre-March advisories to unscheduled.

These are not investments but defensive costs incurred by organizations that do not earn extra revenue from them.

It is crucial to note that there is no measured penalty for AI content. A study by Ahrefs found no correlation between the percentage of AI-written pages and ranking positions, while Google's spam policy remains indifferent to authorship. Anyone claiming that Google prioritizes human-authored content should reconsider their stance.