Read the recap below or watch the recording here.
Setting the stage
Ryan opened by asking each panelist why misinformation feels more urgent today. Suze described technology as a potential equalizer that can open up access to knowledge, while noting that the volume of content now being produced, much of it generated by AI, can look convincing whether or not it is accurate. The challenge, she said, is giving people the tools to tell the difference so they can make better decisions. She also raised the related problem of missing information, pointing to censorship of research as part of the same landscape.Meagan’s concern centered on how misinformation spreads. “Misinformation is very good at storytelling, using elements like people, and emotion, and surprise to make what it’s sharing memorable,” she said, adding that publishers can do more to make the research they publish meaningful and easier for people to find their way into. For David, the stakes are immediate. “As the publisher of The New England Journal of Medicine, obviously our bread and butter is our trust and reputation,” he said, noting that medical misinformation is harming people, changing the relationship between physicians and patients, and politicizing research.
Standards for AI use
Ryan noted that many people, including researchers, don’t realize AI-generated summaries aren’t vetted by humans, and that authors and reviewers are already using AI in their work. David said there are basic principles every publisher should adopt, including disclosure of how AI was used and expectations around accuracy and originality. NEJM follows the International Committee of Medical Journal Editors guidance, but its journal NEJM AI is piloting a different approach because reviewers in that field are hard to find. With author opt-in, papers likely to be accepted receive AI-generated reviews alongside a human review, and human reviewers must finish their own assessment before seeing the AI output. “At the end of the day, the human makes the decision,” David said. His broader point was that policies will need to vary. “Each journal, each field is going to have to have their own policies on AI use,” and those policies will need regular updates as the technology changes.Suze raised a caution about bias, since AI models are trained on historical data in which many groups have been underrepresented, and she was interested in how publishers can avoid reinforcing those patterns.
Telling the story of science
A recurring theme of the session was storytelling, and Suze was quick to separate the word from fairy tales. Storytelling, she said, gives us a way to communicate the values behind research and to show people how it affects their lives. Meagan agreed, pointing out that even content that is fully machine readable, with complete provenance, isn’t necessarily meaningful to a human reader. The Science family has been experimenting with presenting research the way misinformation presents its claims, putting people, values, and emotion at the front without sacrificing rigor. In a recent press briefing on using psilocybin to treat chemotherapy-induced nerve pain, the team asked researchers why they took on the work, and they shared stories of family members lost to opioid addiction. “It changed the conversation,” Meagan said. “Science has all of these elements inherently. Story, people, surprise, emotion.”Suze argued that science communication is too often treated as a nice-to-have, with little recognition in workload models, promotion incentives, or training. “It’s not dumbing down,” she said, describing it instead as finding the right analogy and context. David questioned whether publishers are the right ones to tell those stories. “We’ve ceded that role to the media, to influencers,” he said, and many of those voices share research out of context. He suggested that proxy storytellers could help, since not every researcher is a natural communicator and not every publisher has the resources to invest in that work.
The panel also discussed where conversations about research now happen. Meagan shared the story of an author who answered questions on Reddit about his butterfly research and ended up recruiting a new graduate student, and her team has since developed guidance for authors engaging on Reddit and LinkedIn. David was more skeptical of social media, since algorithms often keep content from reaching the people it’s meant for, and his team is focusing on making sure AI chatbots ground their answers in the best evidence. Suze reminded the room that public trust in scientists remains high, according to several recent surveys, which shapes how we should communicate. “Rather than convincing, sometimes you just want to present,” she said.
Retractions in an AI era
The conversation then turned to what happens when retracted or outdated research becomes part of the data that AI models learn from. Drawing on NEJM’s experience licensing content to AI companies, David said publishers need automated ways to send corrections and retractions to licensees and to make sure those updates reach the data the tools rely on. “A retraction for a machine that really never learns about it in a timely way isn’t really a retraction,” he said. He added that a licensor’s responsibility doesn’t end at signing, and NEJM is now thinking about how to audit the performance of the tools it licenses content to.Suze and Meagan focused on the culture around retractions. Suze pointed to Nobel laureate Frances Arnold, whose openness about retracting a paper helped normalize the process, and Meagan highlighted the first Ctrl-Z Award, which recognizes researchers who retracted their own work at professional risk. Meagan also described the Science family’s “Breaking the Silence” editorial, developed with other publishers, Retraction Watch reporters, university communications officers, and research integrity officers to help authors respond to media questions when concerns about their work surface.
What deserves action now
Ryan closed by borrowing a question from Wendy Queen’s opening keynote and asking each panelist what deserves action now. Suze called for more community engagement with researchers to understand how they’re using AI and other new tools, noting that these technologies can help level the playing field for people who haven’t had the same access to publication. “This is a global endeavor,” she said, encouraging publishers to be partners in change while keeping research integrity at the center.Meagan emphasized balance. “We have to work on the machine-readable piece, but we also have to work on the human-readable piece,” she said, pointing to short summaries and author videos that explain what gap the research fills and why it matters. David urged the room to stay engaged. “Be the person in the arena. Don’t sit on the sidelines, because this isn’t going away,” he said. He sees a responsibility for publishers to make sure peer-reviewed research grounds the answers physicians and patients receive from AI tools, and to find ways to carry the context and human judgment that medicine depends on into those systems.
The session made clear that trust in science is still strong, and that publishers have a meaningful role in protecting it. That work will depend on telling the human story of research, building the infrastructure to keep the scholarly record accurate wherever it travels, and staying engaged with the people and platforms where the public now looks for answers.
View all of the Platform Strategies 2026 session recordings here.