December 2022. In a gleaming office building in Lausanne, Switzerland, the executive team at Frontiers Media had every reason to celebrate. They'd just published their 125,000th article of the year—a staggering 47% increase from 2021.
The numbers told a story of unbridled success. From 54 journals in 2016 to over 230 by 2022. Articles viewed and downloaded 2.1 billion times globally. The third most-cited publisher among the world's 20 largest. By every conventional metric, Frontiers was winning.
But CEO Kamila Markram saw something else in those numbers. Something that kept her up at night.
While Frontiers celebrated its growth, a parallel industry was flourishing in the shadows. Paper mills—sophisticated operations churning out fraudulent research papers for sale to researchers desperate to pad their publication records.
These weren't crude forgeries. Paper mills employed their own writers, manufactured fake data, generated convincing-looking charts, and even created fictitious author identities. Some used AI to generate plausible-sounding abstracts, substituting unusual words to evade plagiarism detection—a technique researchers dubbed "tortured phrases."
Estimated fraudulent papers published annually across major publishers by 2023
In 2023, Wiley retracted over 11,000 papers. Springer Nature pulled thousands more. The crisis wasn't just about individual bad actors—it was industrial-scale fraud threatening the foundation of scientific publishing.
And fast-growing open-access publishers like Frontiers? They were prime targets.
Frontiers faced a choice that would define its future. They could continue growing, implementing modest quality controls while riding the wave of open-access publishing. Or they could do something radical.
In early 2023, Markram made a decision that baffled industry observers: Frontiers would deliberately shrink. Not gradually. Not apologetically. But decisively.
This wasn't a business downturn. It was a strategic pivot. Frontiers would reject more papers, implement stricter screening, and prioritize research integrity over growth metrics. In an industry where "publish or perish" drives behavior at every level, this was heresy.
The weapon in Frontiers' quality war had a name: AIRA (Artificial Intelligence Review Assistant). Launched in 2018 but dramatically enhanced in 2023-2024, AIRA represented something unprecedented—using AI to fight AI-generated fraud.
AIRA scans for plagiarism patterns, image manipulation, statistically impossible data, tortured phrases, and the telltale signatures of paper mill content. But here's what made it remarkable: it learns.
Each detected fraud refines AIRA's algorithms. Each new paper mill technique gets added to its detection capabilities. In 2024, Frontiers integrated additional tools—Cactus' Paperpal Preflight and Clear Skies' Papermill Alarm—creating what they call a "multilayered defense against organized research fraud."
Proportion of 2024 article rejections performed by the research integrity team
Think about that number. More than half of all rejections now come from integrity screening, before peer review. The gatekeeping had fundamentally shifted.
Technology alone couldn't solve this. AIRA flags suspicious submissions, but humans make the final call. This required training an army of editors to spot sophisticated fraud.
In November 2024, Frontiers held webinars on publication fraud prevention. Nearly 1,000 editors participated. They learned to recognize paper mill fingerprints: overly similar writing patterns across supposedly unrelated papers, unusual citation networks, suspicious author affiliations.
The editors weren't just learning to reject bad papers. They were becoming forensic investigators, piecing together evidence of organized fraud networks. In mid-2024, Frontiers' research integrity team uncovered an entire peer review manipulation network—fake reviewers, fabricated credentials, coordinated schemes to get fraudulent papers approved.
This was war, fought one manuscript at a time.
Here's where the story gets really interesting. By every traditional business metric, Frontiers should have been in crisis. Revenue down. Output down. Market share down.
But something unexpected happened. As they rejected more papers and published fewer articles, quality perceptions soared.
Researchers rating article quality as excellent or good (2024 survey)
Researchers rating peer review quality as excellent or good
Published authors rating their submission-to-publication experience positively
These aren't marginal improvements. These are the numbers of a publisher that researchers trust. And in academia, trust is currency.
Despite—or perhaps because of—publishing fewer articles, Frontiers' research impact metrics tell a compelling story.
Take Frontiers in Science, their flagship selective journal launched in February 2023. In 2024, it published just 65 carefully curated articles across 13 research hubs. Not thousands. Sixty-five.
Those 65 articles generated:
That's an average of 13,000 views per article. When you publish quality, people pay attention.
Meanwhile, across all Frontiers journals, articles were viewed and downloaded 950 million times in 2024 alone, bringing the cumulative total to over 3.7 billion.
Perhaps the most telling indicator of Frontiers' strategic success: institutional partnerships.
In 2024, Frontiers finalized "flat fee" publishing agreements with some of the world's most prestigious research institutions:
By year-end, 76 institutional partners had converted to the flat fee model. These aren't vanity partnerships. Universities don't commit institutional funds to publishers they don't trust.
While fighting fraud, Frontiers didn't lose sight of its founding mission: making research accessible to everyone, everywhere.
The Frontiers Planet Prize, focused on urgent environmental challenges, expanded in 2024:
Live events convened 3,200+ scientists and policymakers to translate research into action on climate change, infectious diseases, and green energy.
This is the paradox: by publishing less, Frontiers arguably increased its real-world impact. Quality publications that people actually read, cite, and use to inform policy matter more than vanity metrics.
Frontiers' story is part of a larger reckoning in academic publishing. The open-access revolution promised to democratize knowledge. And it did—brilliantly.
But it also created perverse incentives. When publishers charge per article (Article Processing Charges, or APCs), the business model encourages volume. More articles = more revenue. Quality controls? They slow things down.
In 2023, researchers published a damning analysis showing that Frontiers and similar publishers faced the "$1 billion question"—could they maintain quality while scaling to thousands of articles monthly?
Frontiers answered with action: No, we can't. So we won't try.
The implications ripple far beyond one publisher. If Frontiers can thrive by prioritizing quality over quantity, what does that say about the "publish or perish" culture?
Consider the researcher early in their career, pressured to publish constantly to secure tenure. If flagship journals start valuing fewer, better papers, the incentive structure begins to shift.
Consider the university evaluating faculty. If citation impact and public engagement matter more than raw publication counts, suddenly quality becomes strategic.
Consider the funding agency deciding which research to support. If published papers are known to be rigorously vetted, trust in the scientific enterprise increases.
Before we declare complete victory, let's acknowledge what we don't know.
First, we're still in the early stages of this transition. Frontiers' quality metrics from 2024 are encouraging, but long-term scientific impact takes years to measure. Citation counts, research influence, real-world applications—these unfold over decades, not quarters.
Second, the financial sustainability remains uncertain. Declining revenue from fewer publications needs to be offset by premium positioning and institutional partnerships. Can they maintain this balance? Time will tell.
Paper mills are adapting too. As AIRA gets smarter, so do fraud techniques. This isn't a problem you solve once and declare victory. It's an ongoing battle requiring constant vigilance and innovation.
Third, culture change in academia is glacial. Even if Frontiers succeeds, will other publishers follow? Will universities adjust their tenure criteria? Will researchers embrace publishing less?
These are open questions. But what's not in question is this: Frontiers made a choice, and early results suggest it was the right one.
In business school, they teach you that growth is success. Market share, revenue, scale—these are the metrics that matter. Shrinking is failure.
Frontiers' story suggests something different. Sometimes the bravest strategy is knowing when not to grow. When to pull back, refocus, and rebuild on stronger foundations.
From 125,000 articles in 2022 to roughly 80,000 in 2024. From fastest-growing publisher to most quality-focused. From fighting off fraud at the margins to making integrity central.
This is what counterintuitive victory looks like. Not perfect. Not finished. But pointed in a direction that the scientific community desperately needs.
In an age where AI can generate thousands of plausible-sounding papers per day, the challenge isn't producing more research. It's ensuring what we publish is worth reading.
Frontiers chose to be worth reading.
This narrative is built on data from multiple sources, cross-referenced for accuracy:
This story follows the principles of narrative journalism—using verified facts to construct a compelling story arc. The "Malcolm Gladwell style" emphasizes:
Directly stated in sources: Publication numbers, quality ratings, AIRA capabilities, institutional partnerships, Frontiers Planet Prize participation
Reasonably inferred: Strategic intent behind decisions, emotional context of choices, comparative positioning vs. competitors
Speculative (clearly marked): Long-term implications for academia, cultural shifts in publishing, future sustainability
Interactive charts show publication trends, quality metrics over time, AIRA's expanding capabilities, and comparative impact measures. All data points in visualizations are sourced from verified reports or calculated from official statistics.
Note on Data: All numerical claims in this story are sourced from official Frontiers publications or verified third-party analyses. Where estimates are used (e.g., 2024 publication volume), they are clearly marked and based on available industry data. The narrative interpretation represents one reading of the evidence, with honest acknowledgment of limitations.