The Ghost in the LLM: Why Zero-Click Discovery is a Predatory Paradise

The Mirage of 'Filtered' Quality

There is a dangerous optimism currently circulating in publishing circles suggesting that the rise of zero-click search and AI overviews will naturally "filter" out low-intent readers, leaving journals with a high-trust, elite audience. This is a comforting delusion. In reality, the decoupling of discovery from the Version of Record (VoR) isn't a filter; it’s a cloaking device for the predatory industry.

The danger is simple. When a scholar bypasses the journal’s actual home (the place where you actually spot the warning signs of a hijacked site or a bogus editorial board) they lose their guardrails. We are sprinting toward a world where the "source" is just a summary. Predatory outfits know this. They are the ones most driven to poison the training data that feeds these machines, turning a summary into a tool for deception.

The Weaponization of the Prompt

Predatory journals have always been early adopters of SEO manipulation. In this new era, they are pivoting to AEO (Answer Engine Optimization). While legitimate societies are debating the ethics of AI, predatory houses are flooding the ecosystem with high-volume, keyword-stuffed, open-access garbage designed specifically to be ingested by LLMs.

If a chatbot delivers the answer without a click, the reader never catches the lack of COPE status or the absurd promise of a peer review in two days. The scammers get everything they want. Their lies get baked into the global knowledge graph, legitimized by a slick UI that the user trusts without a second thought.

Fraud at the Speed of Inference

In her recent analysis, Stephanie Lovegrove Hansen suggests that the disappearance of the platform makes organizational brand more important than ever, and while she is right that the "human is the premium," we must acknowledge that humans are remarkably easy to spoof in a digital-first environment.

The "integrity gap" is getting wider as discovery layers snatch up more traffic. Scammers don't need a pretty site anymore. They just need metadata that looks real to a crawler. They are turning into "ghost publishers," existing purely as bits in a data stream. Without cryptographically signed papers or blockchain-backed review trails, we are giving the house keys to whoever can scream the loudest with the most plausible-sounding noise.

The Death of the 'Visual Check'

For years, we told early-career researchers to just look at the site. Check the fonts, the layout, the professional polish. Those days are gone. When info is chopped up into a zero-click list, those clues for truth simply vanish.

To combat this, we need radical structural reform. We must stop measuring "engagement" and start measuring "provenance." I propose a Mandatory Provenance Protocol: Discovery engines must be forced to display an "Integrity Score" alongside any research summary, a score derived not from citations (which are easily gamed), but from verified institutional affiliations and registered peer-review reports.

The Sovereignty of the First-Party Audience

Owned channels are the only real defense left. For societies, this objective isn't just about marketing; it is about the survival of truth itself. Predatory actors can't fake a live conference or a valid member list yet. If we let answer engines take over, we lose juice and the power to spot a total hallucination. The future of ethics isn't better tags. It is about taking back the direct link to the scholar and making the journal site a required destination.