The Integrity Debt: Why CC-BY is the New Trojan Horse for Predatory AI Substrates

The Open Access Mirage: Infrastructure is a Liability

For years, we’ve operated under the comforting delusion that the more open a paper is, the more it serves the progress of science. We championed the Creative Commons (CC-BY) license as a beacon of liberation. But as we sit here in March 2026, it’s time to confront a brutal reality: our commitment to openness has unwittingly built the ultimate high-speed highway for disinformation.

The current state of peer review is a mess. AI isn't making it better, it's just helping people skip the work. While big names like PLOS say that open licenses help AI learn better, they are ignoring the rot. Predatory journals have spent years filling the world's databases with garbage that looks like science. Now, large language models are sucking up that trash and the good stuff at the same time. These machines give a fake clinical trial from a scam journal the same weight as a real study in a top tier publication.

The "Laundering" of Fraudulent Science

We have officially entered the era of the Great Laundering. This isn't just about some obscure academic padding a CV anymore. Today, predatory operations are engineered for the silicon appetite. They strip out the noise, forge their origins, and dump machine-generated nonsense into the ecosystem. The goal is simple: get caught in the dragnet of the next major model's training set.

This isn't just a failure of individual journals; it's a systemic failure of our digital hygiene. As Alison Mudditt rightly notes in her recent piece, What AI Asks of Open Access, the most defensible asset isn't the content, but the authority. However, I would argue that authority is worthless if the machine reading your paper cannot distinguish between a "Version of Record" and a "Version of Fraud."

If the foundation of AI is built on grabbing every CC-BY file in sight, then the bad actors have already won. They are the ones shaping how future machines will reason. We are basically giving the collective scientific mind a lobotomy by letting these fake studies mix with real work. It is a disaster waiting to happen.

The Failure of the Metadata Gatekeepers

The current crisis exposes a catastrophic under-investment in our integrity infrastructure. Crossref and ORCID are fighting a war with 2015 tools against 2026 threats. Predatory outfits are now spoofing Crossmark signals, creating a hall of mirrors where the "machine-readable signals" that legitimate publishers rely on are being forged at scale.

We do not need more open doors, we need a filter. An Aggressive Integrity Layer is the only way forward. Think of it as a guarded gate between the open web and the AI scrapers. If a journal refuses to follow basic ethics or hides its review process, the scrapers shouldn't even know it exists. We have to make the garbage invisible.

Toward a Radical Re-Architecture of Trust

To save the scientific record, we must stop treating "Open" as a synonym for "Good." The future of scholarly publishing isn't the PDF, and it isn't even the CC-BY license. It is the **Verified Provenance Chain**.

The Death of the Anonymous Scrape: AI developers should be held ethically (and perhaps legally) liable for using training data from known predatory outlets. We need a Global Blacklist of non-compliant journals that is hard-coded into the exclusion protocols of every major AI laboratory.

Peer Review as Data: We must stop treating peer review as a hidden process. To survive the AI age, the review itself must be published as structured metadata. If an AI cannot "see" the critique that shaped the paper, it shouldn't be allowed to "trust" the findings.

If we keep giving away the good stuff while scammers flood the zone with high-speed junk, our openness becomes a trap. The dream of including everyone is dead. Now is the time for a much tougher approach to protect the truth. It is a big deal, and we are running out of time.