Rethinking Peer Review in the AI Era (Peer Review Week 2025)

Jenny Bright, our Research Integrity Manager, was fortunate to attend a number of webinars and talks during Peer Review Week 2025. Here, she shares her reflections and key takeaways from the discussions.


AI in Peer Review: Challenge or Opportunity?

The key talking point across the week was artificial intelligence (AI), with a primary focus on “Rethinking Peer Review in the AI Era.”

We discussed how AI is reshaping the peer-review process – and while it presents a number of challenges, it’s important not to overlook the opportunities it brings.

The consensus was clear: AI is here to stay. Publishers must tackle it head-on, implementing robust tools and procedures to identify AI use, ensuring publication integrity is maintained and that AI involvement is made transparent to both authors and readers.

Where AI Is Making an Impact

AI is now prevalent in academic publishing in many areas – from AI-generated articles and manipulated images or data, to the use of AI for conducting reviews, and the tools publishers use to detect and combat AI misuse.

As AI tools become increasingly embedded in academic workflows, questions have been raised around ethics, bias and transparency, such as:

  • How can AI enhance reviewer efficiency without compromising integrity?

  • What ethical frameworks should guide AI-assisted peer review?

  • Which aspects of peer review must remain distinctly human?

My Key Takeaways from Peer Review Week

  • AI in peer review should be a support tool, not a replacement.
    Experts agreed that while AI can streamline tasks like plagiarism detection, citation checking, and reviewer matching; human judgment and critical analysis remain essential for evaluating novelty and scientific rigour. It was agreed that AI and humans together can produce an effective peer-review process.

  • Industry-wide standards are necessary.
    Every publisher is currently handling AI differently, without an overarching set of guidelines. There’s a growing need for clear, cross-publisher policies on how AI should be managed and used throughout the peer-review process, with regular monitoring and review as AI evolves.

  • Combatting papermills and identifying data/image manipulation.
    AI is a positive and invaluable tool to help detect patterns of misconduct, such as AI use, papermill submissions and citation stacking. There are some pioneering tools in development, and used correctly they can identify ethics concerns in a shorter timescale than would be humanly possible, however they must be tested and used responsibly to avoid false positives.

  • Reviewer training and support.
    Institutions must educate reviewers on AI tools and their limitations to ensure informed and ethical use.

  • Preserving trust and transparency.
    Across the week, the publishing community emphasised that trust in peer review is fundamentally important, and there must be openness about how decisions are made – especially when AI is involved.

In Summary

The emergence of AI in peer review, and academic publishing more broadly, is both a threat and an opportunity.

The peer review landscape is changing (and will continue to change) as a result of AI development. We must identify the risks, establish robust safeguards at every potential point of AI integration, and harness AI’s ability to detect anomalies and patterns to combat misconduct.

By doing so, we can ensure research and publications remain high-quality, transparent and trustworthy across scholarly publishing.

From Games to Generative AI: What We Took Home from ALPSP 2025

From Games to Generative AI: What We Took Home from ALPSP 2025

The ALPSP 2025 Annual Conference ran from Wednesday to Friday last week (10-12 September). After a busy few days catching up back in the office, we’ve had time to pause, reflect and draw out some of the key themes.

What struck us most was that innovation in publishing isn’t just about adopting the latest tools – it’s about solving real problems for researchers, editors and communities. Across the sessions, five themes stood out: user-centred design, authentic community engagement, pragmatic use of AI, new approaches to peer review, and smarter digital communications.

Here’s our recap of the highlights:

Session 1: Innovation Lessons from a Viral Tabletop Game

Speaker: Max Hui Bai (Director, Political Belief Lab)

Who would have thought a tabletop game could hold lessons for academic publishing? Max Hui Bai’s talk based on his ‘Publish or Perish’ game highlighted six principles of innovation that publishers can learn from:

  • Understand users deeply – what do they need, and what frustrates them?
  • Contrarian thinking – challenge conventional wisdom to uncover new opportunities.
  • Edge-case thinking – solutions for a niche often benefit the many (think wheelchair ramps → also helping people with pushchairs, deliveries and travel).
  • Design for emotion – people remember how an experience felt, not just whether it worked.
  • Rapid feedback loops – observe real users in action and adapt quickly.
  • Business model innovation – sometimes the biggest change is in how we fund and deliver knowledge.

Takeaway: Academic publishing needs to experiment more boldly, and not just with tech – business models and emotional experience matter too.

Session 2: Bridging the Gap – Community Involvement in Publishing

Speakers: Godwyns Onwuchekwa (Global Tapestry Consulting), Funmilayo Doherty (Yaba College of Technology), Kaitlin Thaney (Invest in Open Infrastructure), Guilherme Werneck (Brazilian Ministry of Health)

The panel stressed that audiences aren’t just consumers – they’re partners. True engagement means:

  • Meeting communities where they are (including outside traditional publishing spaces).
  • Acknowledging and addressing power imbalances.
  • Showing action and responsiveness, not just listening.
  • Starting small, proving impact and scaling gradually.

Takeaway: Engagement is meaningful when it’s built on trust, which must be earned through truly reciprocal partnerships.

Session 3: Awards Finalists Lightning Presentations

Moderator: David Sommer (Kudos)

Some of the most exciting projects spotlighted included:

  • Atom Grants – an AI-powered search engine for identifying suitable funding grants.
  • Frontiers for Young Minds – authors (including Nobel laureates) rewriting their research for 8-15 year-olds, peer-reviewed by kids.
  • AI Talks with Bone & Joint – AI-generated podcasts to engage busy surgeons.
  • Alchemist Review (Hum/Grounded AI) – rapid extraction and summarisation of manuscripts’  key insights (including citation and reference evaluation) to streamline editorial workflows.
  • Taylor & Francis/DataSeer – AI tools to improve data transparency and reproducibility, checking manuscripts for compliance and informing future policy.

Takeaway: AI is already embedded in workflows – but the best examples are those which keep humans in the loop.

Session 4: Innovations in Peer Review

Speakers: Lynsey Haire (Maverick Publishing Specialists), Laura Feetham-Walker (IOP Publishing), Christopher Leonard (Cactus Communications), Enric Sayas (MDPI)

Peer review is under pressure from rising submissions coupled with reviewer scarcity/fatigue. Innovations discussed included:

  • AI in review (MDPI) – tools such as Finder and Ethicality are useful for finding reviewers, spotting ethical issues and quality checks – but require human oversight.
  • Professionalising peer review (Cactus Communications) – could/should we treat it as a recognised profession?
  • Co-review models (IOP Publishing) – involving early-career researchers under supervision to diversify and expand reviewer pools.

Takeaway: Peer review is evolving into a hybrid ecosystem – AI accelerates, but human judgement and trust remain essential.

Session 5a: Digital Engagement and Social Media

Speakers: Kelly Henwood (TBI Communications), Jitske de Vries (The Company of Biologists), Tina Harseim (Springer Nature), Charlie Rapple (Kudos).

Key lessons on social media strategy:

  • Long-form content can work on social, if supported by bite-sized fragments linking back to it.
  • Paid social drives reach and amplifies organic growth.
  • LinkedIn remains important – with staff acting as brand ambassadors outperforming corporate accounts.
  • Newer platforms like Bluesky offer engagement opportunities, while X (Twitter) still drives traffic.

Takeaway: Fragment and reframe content for different audiences and use staff networks as amplifiers.

Session 8a: Portfolio Development in the Age of Volume

Speakers: Kevin Lomangino (KGL Consulting), Sophia Anderton (BJUI), Clare Curtis (Biochemical Society), Gwen Twillman (American Society for Nutrition)

Case studies showed how societies and associations are adapting:

  • BJUI’s partnerships with 35 associations and the development of 450 self-directed learning modules drive users to the journals to extend reach without a member base of its own.
  • ASN outsourced its publishing operations and has invested in a strategic planning initiative which will result in a more coordinated approach across the portfolio.
  • Biochemical Society pivoted away from its young OA journal, instead expanding existing journals’ scope.

Takeaway: Portfolio management is about mission alignment, not just adding titles. Smarter, not necessarily bigger.

Final Thoughts

After a week back in the office catching up on day-to-day work, it’s been valuable to pause and reflect on what ALPSP 2025 offered. The sessions highlighted that innovation isn’t just about new tech (AI!). It’s about rethinking models, respecting communities, and designing for people first.

As we return to business as usual, these lessons remind us to continue shaping our work: being bolder in experimentation, more intentional in community engagement, and mindful in how we balance technology with trust.

 

COPE: Ethics in the Age of AI

COPE Forum 2025: Ethics in the Age of AI

The COPE Forum on 1 July 2025 was a thought-provoking gathering. Dissecting the dilemmas AI brings to the editorial table, the forum spotlighted the evolving challenges – and possibilities – in publication ethics.

Hot Topic: Artificial Intelligence in Publishing

A major theme of the discussion revolved around how AI is reshaping scholarly publishing:

  • Generative AI is influencing everything – from research paper drafting to figure creation and peer review.
  • Concerns were raised around transparency, quality control, and the ethical boundaries of AI-driven content.

The growing consensus? AI is a collaborator, not a co-author. Most policies and guidelines reject AI tools like ChatGPT being listed as authors, however disclosure statements are encouraged to highlight if an AI tool has been used in the creation of the work.

Ethical Guidance from Key Organizations

COPE, along with STM, WAME and EASE, is developing frameworks to guide responsible AI use:

  • Disclosure expectations: Authors must clearly acknowledge any AI assistance in manuscript preparation.
  • Classifying AI use: A helpful chart distinguishes acceptable use (editing and formatting text) from questionable practices (generating data or results).

Editors, Reviewers & Publishers – Stay Human, Stay Honest

AI is showing promise in streamlining workflows for editors and reviewers. Largely with a focus on using AI tools for checking the clarity and quality of language use and to support reference checks within manuscripts.

However, reviewers and editors must validate all content themselves. Uploading confidential papers into AI tools could breach trust and privacy and compromise the Publisher by sharing third party data.

The LOCAD Framework: Vetting AI Tools

Before using any AI tool, it is advisable to apply the LOCAD test:

  • Limitations: Know the tool’s scope.
  • Ownership: Understand rights over the content it generates.
  • Confidentiality: Protect sensitive data.
  • Accuracy: Validate the information.
  • Disclosure: Be upfront about usage.

Final Thoughts: The Future Is Collaborative

The forum concluded on a hopeful note:

“AI should work with us, not instead of us.”

Ethical governance, responsible use, and human creativity must remain the core of scholarly publishing – even as AI becomes a more integrated partner in the process.

Text recycling in research writing (COPE Lightning Talk, April 2024)

The “text recycling in research writing” COPE Lightning Talk in April 2024 was presented by Professor Cary Moskovitz (Director of the Text Recycling Research Project at Duke University, NC, USA).

 

Text recycling can be defined a number of ways. Essentially it is the reuse of material – whether it is prose, visuals or equations – in a new document where material is used in the new document as it was in the original source. The material is not presented in the new document as a quotation and usually one author on the new document is the same as the original.

 

Recycling material in published papers is common, however authors do need to be mindful of the copyright law that they are submitting their paper under. Under US copyright law, authors may find that generally they can reuse portions of text under the Fair Use clause, and in many STEM journals publishers ensure their policies work to accommodate this, so as not to infringe the copyright/legal ownership.

 

The ethics of recycling are different and it requires appropriate transparency. Legalities and ethics may differ and the author must ensure that permission has been requested if their use of copyrighted material falls outside of the Fair Use guidance and journal policies. Text recycling should be done within best practice – we cannot expect authors and editors to be experts in copyright law, however publishers’ policies should adapt to and support their communities. It is also important to highlight that authors should check their publisher agreements as this is where the details lie. These will vary from publisher to publisher and the rules around text recycling are not always consistent.

 

It is recommended to authors that if they are using work which is previously published, a statement could be included in the paper to highlight this, and this would act in accordance with best practice. This will allow the reader to know that the work is not original, but it will also signpost accordingly.

Ethical considerations around watchlists (COPE Forum, March 2024)

The March 2024 COPE Forum began with a discussion on ‘ethical considerations around watchlists’, hosted by Daniel Kulp (Chair, COPE Council).

During the discussion, the benefits and potential risks associated with watchlists were explored from the position of both individuals and journals, as well as how publishers should use them. In principle, a watchlist could be used as a quality control tool; usually kept private as a measure of accountability. If, however, the list was exposed/leaked, the publisher’s trustworthiness and maker of transparency can be held to account.

It is important to note what is included on a watchlist – is information recorded about an individual, including personal data? Do COPE offer any guidance on what kind of criteria individuals should meet to be on a watchlist? And how can an individual defend themselves or “remove” themselves from a list? These were some of the key questions raised during the conversation regarding individual authors or account holders. The threat is much greater to be named and listed; as careers, research programs and scholarly publications can all be potentially be disputed/disrupted if this extremely sensitive information is exposed in the public domain.

At an individual level, abhorrent practices to circumnavigate peer-review processes could be observed through “flagging” an author’s account – for example, to collate information about suspicious activity. However individuals have a higher right to privacy and such lists need to be regulated or kept private within the publishing house. Academic publishing is witnessing a systematic effort to undermine peer review; the rise of papermills, reviewer cartels and bad actors exploiting the system often for financial gain mean that individuals should be aware of this and make due considerations during submission.

Sharing nefarious behaviours is important and if publishers share systematic problems regarding behaviour amongst one another, they can collectively disrupt the ways papermills and peer-review rings operate. COPE has guidance on how EICs can share information. However, how can we share this information more broadly? The STM Integrity Hub offers a duplicate submissions check across publishers, which could be a useful tool to manage misconduct. This is being developed and the STM Integrity Hub are exploring managing this in an ethical and legal way.

One of the biggest risks with watchlists posed at an individual level is that account holders are added incorrectly or without proper reason. This could be very detrimental to an individual in terms of their career and may also possibly be defamatory and could have legal ramifications. Watchlists must aim to be transparent but not give away details of what we are looking for. The risk is higher for individuals as it can directly impact future employment and ability to do research. Should there be mitigation circumstances for appeal for minor infractions? A similar name could be a risk, for example.

In the absence of an agreed decision amongst the industry on what should and shouldn’t be on a watchlist, the main risk appears to be a legal one. STM have already advised that the industry needs to be careful regarding the type of data which is shared. Legally underpinning this stage is crucial to ensure an individual’s privacy and GDPR-compliance is maintained. It was suggested that if private watchlists are maintained, legal counsel should be able to advise how to evoke sanctions or prevent someone from publishing. The publisher however must be mindful of any reputational damage or litigation brought against them.

If an individual finds themselves on a watchlist they do of course have the right to reply, to ensure that the process is fair and transparent and that they can defend themselves and their position. It was raised that it would be good practice to notify an individual that concerns have been identified, which will allow them to ask for an appeal. It is important to consider and understand what the watchlist represents.

As the risks increase and the systematic manipulation of peer review is attacked, the ways in which this data is captured in order to identify patterns of misconduct has to be carried out quickly, ethically and legally in order for the publisher to maintain integrity and be commercially viable as well as managing its own trustworthiness.

The Editorial Hub at the London Book Fair 2024

On Tuesday 12th March we headed off to one of the industry’s leading events in academic publishing. This year the London Book Fair was buzzing; the main hall was filling up as soon as the doors opened, and the Tech Theatre and Main Stages had consistent queues.  Meetings, networking and seminars are all key elements of the fair and this year we were lucky enough to participate and find out more about the challenges facing the industry. 

 

 

The hot topic as ever was AI, and we attended a talk hosted by Sureshkumar Parandhaman (AVP of Publishing Solutions and Pre-Sales, Integra) entitled “Embracing AI in Publishing: Transform Editorial Excellence and Enhance Downstream Efficiency”. Needless to say it’s evident that tools are being developed to tackle many of the administrative tasks that editorial services provide, however we know there will still be an important role for us to play as integrity, quality and efficiency are all integral within our roles.

 

During a key discussion about AI and copyright it was stated that when King Charles’ recent surgery was announced, over 250 AI-generated texts filled the virtual Amazon bookshelves the very next day! Content is quick to make, and the session “Global Discussion of Machines, Humans, and the Law” hosted by Glenn Rollans (President and Publisher, Brush Education Inc.), Nicola Solomon (CEO, Society of Authors), Porter Anderson (Editor-in-Chief, Publishing Perspectives), Dan Conway (CEO, The Publishers Association UK) and Maria A. Pallante (President and CEO, Association of American Publishers) highlighted how important it is for the copyright holder to be accredited and financially rewarded. Whilst Government bodies manage policy and next steps with regards to AI, we need to be mindful of what we “feed” to LLMs and how “Fair Use” comes into play.

 

The Editorial Hub at LBF 2024

A panel discussion entitled “AI & Publishing – Navigating the Impact of Large Language Models” was hosted by Lucy McCarraher (Founder, Rethink Press Limited), Nadim Sadek (Founder & CEO, Shimmr AI Ltd), and Sara Lloyd (Group Communications Director & Global AI Lead, Pan Macmillan). The group discussed the ways that LLMs can spark creative thinking and should be used within workshops as a tool to support creativity. In terms of optimising the features of AI it was recommended that the industry looks to use it for creating the structural analysis needed for promotion work. An AI bot can extract book DNA far quicker than a pair of human eyes, so let’s use its skillset appropriately and it can enhance our work, rather than hinder progress. As well as creating copy, it can also be useful for generating advertising and images. As well as connecting couriers and customers, it can scroll databases to highlight potential consumers. The panel really encouraged the audience to engage with the features of AI bots and embrace how useful it can be to enhance and build on capabilities.