The Cannibalistic Record: Why ‘Plausible Fraud’ is the New Business Model

The Great Dilution Has Begun

Peer review isn’t being challenged by AI; it is being bypassed by a wholesale industrialization of "plausible fraud." For decades, we have operated under the comfortable delusion that scholarly publishing is a gatekeeping exercise. We believed that if a paper looked like a paper, cited like a paper, and smelled like a paper, it was the result of a human sweating over a microscope.

That era ended the moment Generative AI achieved a passable scientific prose style. We aren't just fighting the occasional rogue researcher anymore. We are facing a total structural collapse. The cost of generating a fake paper has dropped to basically zero, while the cost of verifying it remains huge. The result is a market flooded with zombie science that looks alive but lacks a soul, or a data set.

The Investigator: Follow the Incentives to the Grave

The rot in our world isn't really the technology (it's the money). We've built a system where publishers have every reason to look the other way. Back in June 2023 at the Society for Scholarly Publishing, Tim Vines and Jessica Miles went at it over whether AI would kill the industry. Everyone talked about detection tools, but they ignored the big deal: for predatory journals, AI content is actually a high margin product.

If you are an author-pays (APC) journal, every submission is a potential check. If that submission is a perfectly formatted, AI-generated hallucination that passes a cursory (and often automated) peer review, why would a profit-driven entity reject it? The "arms race" is a myth because one side has already defected to the enemy. The predatory journals of 2023 are no longer just clunky operations in backrooms; they are becoming AI-optimized carousels, turning prompt-engineered manuscripts into line items on a CV for a fee.

The Library of Babel is Real

We are hitting a point where the sheer volume of synthetic crap outweighs the actual scientific record. It is a death loop. As Vines noted, if we train the next generation of AI research tools on a poisoned record, the output becomes a hall of mirrors. We are basically feeding the future of science a diet of high fructose nonsense.

Traditional publishers claim that "trust and transparency" will save them, citing investments in plagiarism tools. This is like bringing a magnifying glass to a forest fire. Current tools are designed to catch *copied* text. AI produces *original* text that is factually vacant. The "Integrity Sentinel" cannot guard a gate that doesn't exist.

Structural Reform: The Death of the 'Paper' as Currency

If we want to survive the next five years, we must kill the manuscript as the primary unit of credit. We need to move toward a "Hard-Object Protocol."

Mandatory Raw Data Sovereignty: If the raw, timestamped data and the executable code aren't deposited in a sovereign, third party repository before the paper is written, the paper does not exist. No data, no DOI. Period.

The Identity Bounty: We must move beyond ORCID's polite suggestions. We need a cryptographic verification of authorship tied to institutional biometrics. If we cannot prove a human was at the lab bench, we must assume a machine was at the prompt.

The industry is currently obsessed with whether AI can help us write. That’s a distraction. The real question is whether we have the courage to stop rewarding people for simply filling a PDF. If we don't change the currency of academia from formatted prose to verifiable evidence, the scholarly record will become nothing more than a high priced scrapheap of synthetic noise.