At Frontiers, research integrity and quality sit at the heart of everything we publish. Supporting this commitment is our dedicated Research Integrity team, working behind the scenes to safeguard the scientific record at scale.

In her latest blog, Elena Vicario, Director of Research Integrity, reflects on 2025 - a demanding but transformative year that reshaped how the team addresses emerging challenges, from papermills and fast-churn science to the responsible use of AI, and how this work is laying a stronger foundation for trust in high-quality science.

As the first month of 2026 comes to an end, I’ve found myself doing something that doesn’t always come naturally in research integrity: pausing to reflect.

Because 2025 was intense. Demanding. At times challenging. Yet ultimately one of the most impactful and rewarding years of my career.

Research integrity rarely slows down long enough for reflection. Reactive cases need immediate attention, and when proactive systems are working well, they’re mostly invisible. But this year felt worth looking back on. Not because the challenges were fewer, but because of the deliberate way we combined planning with action.

We didn’t just respond to problems. We tried to get ahead of them.

Taking the fight to papermills at scale

Papermills are not new. What has changed is their scale and coordination, and how effectively they exploit gaps between systems, journals, and publishers.

In 2025, we strengthened our approach by assessing industry AI tools for papermill detection and integrating two alongside our own quality checks in AIRA. The goal was never to rely on a single signal. It was about patterns, context, and corroboration, and about giving teams better evidence to make difficult calls earlier.

This moved us from isolated checks to a layered, more defensible integrity framework. Detection works best when tools reinforce each other rather than operate in silos.

Fast-churn science: naming the problem and raising the bar

Another focus was what we began calling fast-churn science.

These are studies that technically meet publication requirements but rely on rapid dataset reuse, limited analytical depth, and production processes optimised for speed and volume rather than real contribution. Left unchecked, they erode trust in the literature.

Instead of treating this as a vague quality concern, we defined it clearly and set firmer expectations. Clearer policies. Clearer guidance. A higher bar for meaningful contribution.

We became the first publisher to require independent validation for these studies and are now sharing what we’ve learned with the wider community to help shape how quality is assessed in an increasingly AI-enabled environment.

Leading responsibly in the AI era

Our AI whitepaper made something clear. AI-assisted science is already everywhere across design, analysis, writing, and peer review. Whether acknowledged or not, AI is now part of the research ecosystem.

That makes trust essential.

If AI influences how research is produced and evaluated, then how it is used and how openly it is discussed matters. Transparency is not optional. It is foundational.

At Frontiers, that means being clear about where AI supports our workflows, what it can and cannot do, and where human judgement remains central.

That is why we chose to give AIRA a voice. Built on more than 16 years of data, it has long supported our Research Integrity teams behind the scenes. Explaining what it does and what it doesn’t was not about promotion. It was about accountability. Human expertise remains at the centre and always will.

Integrity does not work in silos

One lesson 2025 reinforced is that research integrity cannot be tackled alone.

We collaborated more closely with other publishers, sharing signals, experiences, and lessons learned. Siloed approaches only help those trying to game the system. Misconduct networks do not respect publisher boundaries, and our responses should not either.

This included investigating a large-scale peer review manipulation network that affected multiple publishers. Addressing it required time, persistence, and genuine cooperation. Sharing the findings publicly was about transparency and prevention, not visibility.

We also engaged more with research integrity sleuths. These conversations were not always easy, and we did not always agree, but they were valuable. They challenged our assumptions and strengthened our investigations. While perspectives differ, the goal is the same: protecting the scientific record.

Putting experts at the centre

Behind every policy and decision are people.

In 2025, we made a deliberate effort to highlight the teams doing this work every day.

Research integrity is not a checklist. The hardest calls depend on judgement, nuance, and careful reasoning under pressure. The Frontiers Research Integrity team carries that responsibility daily, driven not by recognition but by a commitment to safeguarding the record.

Making that work more visible mattered. When only outcomes are seen and not the thinking behind them, trust is harder to build. Showing the care and reasoning behind decisions helps create more constructive conversations.

Looking ahead

What 2025 made clear is that research integrity is no longer about isolated cases at the margins. The challenges are systemic, shaped by incentives, scale, and rapidly evolving behaviours across the ecosystem.

Meeting them requires approaches that are robust, fair, and transparent by design.

The work ahead is substantial. But the dedication and judgement I’ve seen over the past year give me real confidence, not just in what we are prepared to face, but in what we can help shape.

For us, 2026 is not just about being ready. It is about setting the direction forward.