Fighting Back to Protect the Scientific Record

www.tonyhopedale.com

Those of us who work in scholarly publishing and research communication are not bystanders to science and scholarship, we are part of the research infrastructure that helps scientific evidence survive. With that in mind, I summarize below a recent webinar that I attended where an interesting and useful new tool, developed by the Union of Concerned Scientists, for tracking attacks on the scientific record was demonstrated and explained. This tool is publicly available, which opens up some interesting opportunities for publishers and science communicators!

In my June 15 post I addressed the Office of Management and Budget (OMB) proposed rule changes for managing U.S. government grants by federal agencies. These rule changes encourage political control over grant review, create barriers to international collaboration, and place restrictions on the use of funds for conference attendance, professional society membership, and publication costs. Much has been written about this in many of the publications and blogs that I follow — I’ve included links to a few recent Scholarly Kitchen highlighting different perspectives on the OMB situation.

I also list a few of the most interesting articles and posts I’ve recently read, including a social media survey from the Society for Scholarly Publishing and a few items spanning the topic of artificial intelligence in research and in publishing. Plus a plug for me new venture, Scholarly Publishing Solutions.

The Attack on Science Is an Attack on the Scientific Record

On July 1, the Union of Concerned Scientists (UCS) hosted a webinar introducing its new *Attacks on Science Tracker, an interactive tool documenting attacks on science and potential scientific integrity violations under the second Trump administration. The purpose of the webinar was to demonstrate the Attacks on Science Tracker* dashboard, but is also served as a warning about what happens when political power suppresses, distorts, or intimidates a nation’s scientific infrastructure.

The Combat of the Giaour and Hassan, Eugène Delacroix, 1826; photo taken at the Art Institute of Chicago

During the webinar attacks on science were described as actions that weaken the role of independent science in federal decision-making, things like politicized or restricted research funding, fired or sidelined scientists, censorship or distortion of scientific information, and the use of misinformation for political purposes. Andrés Bachelet of UCS emphasized that these actions matter because they undermine the science used to protect public health, safety, the environment, and communities.

What is the Attack on Science Tracker?

Dr. Jules Barbati-Dajches explained that the Attacks on Science Tracker is a public, interactive tool that documents actions weakening the role of science in federal decision-making. Users can explore attacks by topic, agency, type of attack, and time period, and can also view the underlying data and methodology. In broad terms, the Tracker turns hundreds of scattered incidents into a searchable record that researchers, advocates, journalists, and policymakers can use to understand patterns and explain why scientific integrity protections matter.

Around the time of the webinar, the Tracker showed 576 attacks on science and 188 potential scientific integrity violations between January 20, 2025 and the end of June 2026. UCS has tracked attacks on federal science since 2001. What is shocking is that the current pace and breadth of this interference has required a more sophisticated, public, interactive tool. The public Tracker, the methodology they use, and the supporting data can be found on the UCS website, including a downloadable dataset and documentation.

In scholarly publishing we often talk about research integrity in terms of plagiarism, image manipulation, paper mills, disclosure failures, and peer review abuse. Those are certainly real threats. But the UCS webinar focused on another aspect: scientific integrity also depends on whether scientists are free to ask questions, collect data, report results, maintain public datasets, communicate findings, and participate in policymaking without retaliation. Publishers can evaluate and correct submitted and published articles; but they cannot easily correct the disappearance of the data, the intimidation of the scientist, or the political alteration of a federal report before the scholarly community ever sees it.

This is where the Tracker plays an useful role. The Tracker is extremely detailed and pretty cool. It allows users to explore attacks by topic, type, time, agency, and completion stage. It can show, for example, whether attacks cluster around health and safety, environment, climate science, energy, equity, or other policy domains; whether those attacks involve data collection, data accessibility, censorship, funding politicization, or altered results; and which federal agencies are implicated. During the webinar’s Q&A, UCS emphasized that each documented attack includes citations, and that human reviewers verify and categorize reports before they enter the dataset. For publishers, editors, and librarians, this is familiar territory: claims need provenance; evidence needs context; data need documentation; reuse requires trust.

The Politics

March for Science, April 2017; photo taken at Boston Common

One of the biggest problems this webinar highlighted for me was that scientific independence in the U.S. is easily threatened by overreaching executive action. The “Gold Standard Science” executive order, something I discussed in my last blog post, is an overtly political move. Masquerading as a series of common-sense statements, when you read between the lines, the gist is that “Gold Standard Science” is whatever the current U.S. administration says it is.

This is where the Scientific Integrity Act becomes central. Introduced in the U.S. House of Representatives in February of 2025, the bill proposes that policies must:

1. ensure that scientific conclusions are not based on political considerations; 2. prohibit the suppression or alteration of findings; and 3. permit researchers to disseminate their findings and engage with the scientific community as appropriate. Scientific integrity policies must also include a mechanism through which stakeholders may report violations of the policy.

This Act is a way to move protections from executive discretion into law. The bill would help ensure that federal science remains free from political interference, require science agencies to adopt clear scientific integrity policies, and protect scientists from retaliation for sharing findings.

How Scholarly Publishing can Help

One of the most exciting opportunities the webinar presented was the use of the Tracker as a “story starter”. Data alone rarely moves policy and stories alone are often dismissed as anecdote. But documented cases and detectable patterns of abuse can turn scattered incidents into a meaningful story, with evidence and facts.

That is where scholarly publishers, societies, editors, and science communicators have a chance to contribute. A society publisher can use the Tracker to identify attacks affecting its discipline and provide the story (backed by data) to its editorial board or public policy committee. A journal editor can commission an editorial explaining how suppressed federal data affects clinical practice, environmental monitoring, public health, or research reproducibility. A librarian can use the dataset to teach students how political interference affects the information supply chain. A publisher can invite authors to explain what happens when grants are politicized or federal advisory boards are disbanded.

Here is where I get on my usual hobby horse regarding interoperability, open data and PIDs. If the integrity of the scientific record depends on evidence, then publishing organizations should strengthen the systems that make evidence durable: rich metadata, persistent identifiers, funder information, data availability statements, versioning, corrections, retractions, preservation, and clear links between policy documents, datasets, preprints, peer-reviewed articles, and public communications. The attacks UCS is tracking often target the core of science: data collection, data accessibility, funding, advisory committees, and the ability of scientists to communicate. Scholarly publishing is one of the places where that core can be strengthened.

Call to Action

First, use the Tracker. Explore the public Attacks on Science Tracker and its methodology and supporting data. Remember, it’s not a static report, but rather an evolving dataset that can support discipline-specific analysis, editorials, society statements, and public education.

Second, support durable scientific integrity protections. Track H.R. 1106and S. 4545, and urge representatives and senators to co-sponsor and support the Scientific Integrity Act.

Third, let’s bring the scholarly publishing community into the public conversation. UCS is organizing scientists and science supporters through Science Rising, which describes itself as a UCS initiative to protect science, justice, integrity, and democracy. UCS also provides resources for an August recess meetings with legislators and a Senate co-sponsorship ask.

Finally, publish, cite, preserve, and speak. The scholarly communication community has tools that authoritarian politics fears: documentation, attribution, peer scrutiny, metadata, public access, versioning, and archived records. We should use them. We should help scientists tell the stories behind the data. We should make attacks on science visible in the fields they damage. And we should remind policymakers that scientific integrity is not a partisan issue. It is part of the infrastructure of democracy.

RECENT READS

Recent SK Articles on the OMB Situation

**Ask the Chefs: How Are You Responding to the OMB Proposed Rule Changes for US Research Funding?**

Lettie Y. Conrad, Robert Harington, Haseeb Irfanullah, Roy Kaufman, Alice Meadows, *The Scholarly Kitchen*, July 7, 2026

**Where Do We Go From Here? How Scientific Societies are Thinking about the Proposed OMB Funding Rule**

Jessica Miles, *The Scholarly Kitchen*, July 8, 2026

**Why Research Libraries Oppose the OMB Revisions to the Uniform Guidance**

Marcel LaFlamme, *The Scholarly Kitchen*, July 9, 2026

SSP Pulse Check Report: Social Media in Scholarly Communications

**Society for Scholarly Publishing, July 1, 2026**

The recent SSP Pulse Check Report looks at the use of social media by its membership and affiliates (e.g., readers of The Scholarly Kitchen and others in scholarly publishing). The results of this poll are interesting to look at and provide useful insight into what the community is thinking and feeling about social media along with how it is being used. As can be expected, the days of Twitter (X) dominance are over, with usage of LinkedIn, Facebook, Bluesky and Instagram outpacing Twitter. Key findings from the Social Media report are below.

Social media has become an increasingly central feature of how scholarly communications organizations build visibility, share research, and connect with their communities. Yet as platforms multiply, fragment, and shift in character, organizations face mounting uncertainty about where to invest limited resources, how to set realistic expectations, and how to evaluate whether their efforts are working. The June 2026 Scholarly Communications Pulse Check surveyed professionals across the ecosystem to take stock of current practice.

Key Findings

LinkedIn dominates, but the platform landscape is diversifying. LinkedIn is the clear professional default followed by Facebook, Bluesky, and Instagram.

Visibility and dissemination are the dominant goals. Organizations primarily use social media for brand visibility and awareness and dissemination of research.

Social media is seen as effective for reach, less so for results. Respondents rate social media highly for brand awareness and content dissemination, but confidence drops for activities tied to tangible outcomes.

Measurement practices are uneven. Engagement metrics including likes, shares, and comments dominate evaluation tactics, while 21% of organizations report no formal measurement at all.

Organic social media is growing; paid remains a niche tactic. More than half of respondents report increased organic social media activity over the past two to three years.

You can access the full report: *Social Media in Scholarly Communications*

RECENT READS ON AI

General-purpose large language models outperform specialized clinical AI tools on medical benchmarks

Krithik Vishwanath, Anton Alyakin, Mrigayu Ghosh, Ali Hage, et. al., *Nature Medicine*, June 12, 2026

Specialized clinical artificial intelligence (AI) tools are entering medical practice at scale. These proprietary large language model (LLM)-based tools promise superior clinical performance to general-purpose frontier LLMs as a result of domain-specific training or retrieval-augmented generation (RAG). Yet, their architectures, base models and training pipelines are not public. Clinicians and health systems must therefore assess their value and safety without independent evidence. Conversely, large training corpora and extensive alignment of frontier LLMs may enable them to challenge clinical AI tools without domain-specific modification. We test this hypothesis by comparing clinical AI tools (OpenEvidence and UpToDate Expert AI) to leading general-purpose LLMs (OpenAI GPT-5.2, Google Gemini 3.1 Pro Preview and Anthropic Claude Opus 4.6). Later, we include auto-enabled Google Search AI Overview as a real-world control frequently encountered by physicians.

General-purpose LLMs outperformed clinical AI tools on the MedQA questions. Among frontier LLMs, Gemini achieved the highest accuracy at 97.4% (95% confidence interval (CI) 95.6%–98.5%), followed by GPT at 94.2% (91.8%–95.9%) and Claude at 90.2% (87.3%–92.5%). Clinical tools scored lower, with OpenEvidence achieving an accuracy of 89.6% (86.6%–92.0%) and UpToDate achieving 88.4% (85.3%–90.9%). Gemini outperformed all other models. GPT outperformed OpenEvidence and Claude.

Making AI Use of Scholarly Content Traceable, Measurable, and Trustworthy: A Meeting Report from Cambridge Scholarly AI workshop

Todd A Carpenter, Tasha Mellins-Cohen, Monica Westin, *The Scholarly Kitchen*, June 25, 2026

Foundationally, the meeting began with the recognition that traditional usage metrics, including downloads, views, searches, and requests, do not fully capture how AI systems interact with scholarly content. AI tools may retrieve passages, retrieve metadata, generate summaries, or synthesize findings across numerous papers without a researcher ever clicking through to the original article or book chapter. Many of the sources that are included in an output may not even be attributed or referenced, for that researcher to follow back if they wanted to.

You can access the full report: *Standards for Provenance, Attribution, and Usage in the AI Ecosystem*

The twilight of the chatbots

Ethan Mollick, *One Useful Thing*, June 30, 2026

Long-running, smart, and self-correcting AI systems do not need constant human intervention, and they require a different way of working. And, as opposed to chatbots, agents come with extra machinery: harnesses that give the AI access to tools and an environment to act in, and apps built for agents like Claude Code or OpenAI’s Codex. As a result, the already increasing ability of AI models can be improved still further by a good harness or app.

So work is increasingly about assigning work to agents, rather than working together with chatbots. A joint study by OpenAI and academic economists shows how quickly this is happening inside their own organization. Critically, it isn’t just coders who are using agents. Legal, HR, and other non-tech functions have adopted agents at nearly the same rate. OpenAI may be a sort of canary in the coal mine for what will happen elsewhere in work.

Are researchers becoming an unpaid workforce for the AI industry?

Graham Kendall, *Publishing with Integrity*, July 1, 2026

Expertise Has Economic Value

If AI systems learn from this material, they are not simply learning scientific facts. They are learning how experts think.

That distinction matters because expertise has enormous commercial value. The world’s leading AI companies are investing billions of dollars to build systems that increasingly perform tasks requiring human judgement. If researchers’ expertise contributes to improving those systems, an important question follows. Are researchers still volunteering their time solely for the benefit of scholarship, or are they also becoming part of an unpaid workforce helping to create commercially valuable AI products?

First impressions of Claude Science

**Christopher Thorpe, LinkedIn, July 9, 2026**

It went further than the literature review and generated an action plan with options for next steps. I asked it to review two of the proposed next steps; it provided detailed execution plans for both, which I then approved. Without any input from me other than approving the plan, it spun up computational environments and executed them. It was a very different experience, and the distance between having a scientific hypothesis and executing some preliminary computational work felt vertiginously short. These were things I’d wanted to do for a while, but merely hadn’t had the time to set them up.

ADVERTISEMENT

**Scholarly Publishing Solutions is a consulting team with decades of experience leading, growing, and modernizing scholarly publishing organizations. Founded by Jennifer Deyton and Tony Alves, SPS focuses on peer review and editorial workflows, including policy development, best practices, systems design, research integrity, workflow improvement, and the practical use of AI.**

Jenn and Tony bring complementary expertise across the operational, strategic, and technology sides of scholarly publishing. Jenn offers deep experience in editorial operations, society publishing, scalable service delivery, and organizational growth, including the discipline that comes from building a small editorial services operation into a mature, successful company. Tony brings broad product and systems experience across the publishing stack, including interoperability, platform strategy, standards development, AI implementation, and research integrity tooling ecosystems.

Together, they help publishers understand what is working, identify what needs to change, and design practical solutions that can be implemented in real-world editorial and publishing environments. Their approach helps organizations improve efficiency, strengthen trust, support sustainable growth, and modernize workflows without losing sight of the values and relationships that make scholarly communication work.

**Contact us at info@ScholarlyPublishingSolutions.com**

*Disclosure: All original content has been compiled, analyzed, and edited by me. AI has been used to organize my notes, compose parts of early drafts, and to copyedit.

Copyright 2026, Tony Alves, Scholarly Publishing Professional. All Rights Reserved.