Academic writing in the age of AI: what research‑led organizations need to know 

 

Artificial intelligence is rapidly becoming part of the research workflow — from drafting and editing manuscripts to supporting literature reviews and data analysis.  

For research‑led organizations, this shift brings both opportunity and responsibility. 

While AI tools can improve efficiency and accessibility, their use raises critical questions around research integrity, authorship, data confidentiality and disclosure. As editorial policies continue to evolve, institutions, funders and R&D leaders are increasingly responsible for ensuring their researchers are supported with clear guidance, trusted training and governance frameworks that protect both scientific standards and organizational reputation. 

To explore how organizations can support responsible AI use, Nature Masterclasses convened a panel on AI in academic writing and publishing with experts from across research, publishing and policy. 

• Kamar Ameen‑Ali is a Senior Lecturer in Biomedical Sciences and Course Leader for BSc (Hons) Biomedical Science at Teesside University, UK. Her research uses human post‑mortem brain tissue to study how neuroinflammation contributes to neurodegenerative diseases and dementia. 
• Sadra Bakhshandeh is a Senior Editor at Nature Reviews Bioengineering and Senior Consulting Editor for Nature Biomedical Engineering, overseeing review and primary research content across the bioengineering field. 
• Ellie Gendle is Head of Journals Policy and Research Integrity at Springer Nature, leading the development, implementation, and governance of editorial policies across the journal portfolio. 
•  J. Wynand Lambrechts is a Principal Electronic Engineer specializing in semiconductors at Incomar Aeronautics, and a part‑time Research Associate at the University of Johannesburg, South Africa. 

Find the webinar recording here and some of the main insights below. 

 

Why responsible AI use is now a leadership issue

As AI tools increasingly shape scientific writing and publishing, academic institutions face growing pressure to ensure their use does not compromise research integrity, confidentiality or trust with publishers and funders. Researchers themselves remain divided on what constitutes acceptable AI use, making institutional guidance more important than ever. 

The following three insights about AI in academic writing can be leveraged by academic institutions to better support their research: 

1) AI should supplement scientific writing, not substitute it 

The panel experts recommend that researchers avoid using AI to draft manuscripts. Instead, researchers should only use AI to edit or refine text they’ve already written.  

For organizations, this reinforces the need to support researchers with training that protects authorship, accountability and scientific voice, and empowers researchers to write with confidence — especially as AI use scales across teams. 

• AI cannot be an author. According to Springer Nature’s editorial policies, Large Language Models (LLMs) cannot take accountability for the work and do not satisfy authorship criteria. 
• Writing preserves a researcher’s voice and intention. AI-generated text often lacks a compelling narrative or analytical depth. Early-career researchers need to build strong scientific writing skills now, as they shape how effectively they communicate their work throughout their careers. 

2) Prompt with caution 

AI tools are still imperfect. They can ‘hallucinate’ references, make false claims and provide biased or inaccurate information. Many AI systems also lack sufficient data security and protection.  

At an institutional level, this highlights the importance of clear guidance around data privacy, confidentiality and acceptable AI use in research workflows.  

Researchers need to: 

• Craft the right prompt. Researchers should be specific about what they’re asking an AI tool to do. Providing detailed goals, context, sources, expectations and examples will help ensure high-quality outputs. 
• Minimize privacy risks by avoiding the inclusion of sensitive, confidential, or copyrighted information in their prompts. 
• Carefully verify any AI-assisted outputs for errors and biases. 

 3) Check individual journal policies on permitted use and required disclosure 

Editorial AI policies are constantly evolving and can vary widely across different journals and publishers. Before submitting a manuscript, researchers should review individual journal policies and disclose any AI use transparently. 

For research‑led organizations, staying aligned with evolving editorial policies is critical to maintaining trust, avoiding rework and protecting institutional reputation. 

Nature Portfolio’s current editorial policy requires AI use to be “properly documented” in the Methods section. The webinar explores what responsible disclosure should entail in greater depth, but it’s generally advisable to over-disclose, rather than under-disclose, any AI use. 

 

What this means for research‑led organizations

Taken together, these insights underline a broader shift: responsible AI use is no longer just an individual researcher’s concern, but an organizational priority. Institutions that proactively invest in guidance, training and expert‑led dialogue will be better positioned to enable innovation while safeguarding research integrity and trust. 

 

Supporting responsible AI use at scale

As AI adoption accelerates across research environments, the challenge for organizations is no longer whether researchers will use these tools, but how to support their use responsibly and consistently. 

Clear guidance on acceptable AI use, transparent disclosure practices and investment in high‑quality training are becoming essential to maintaining trust with publishers, funders and the wider scientific community. Organizations that take a proactive approach — aligning researcher support with established editorial standards and expert insight — will be better positioned to safeguard research integrity while enabling innovation. 

 

Partner with Nature to support responsible AI use in research 

Nature Partnerships works with research‑led organizations across academia, industry and healthcare to deliver trusted content, training and engagement programs that support researchers — while protecting scientific integrity and organizational reputation. 

From expert‑led webinars and editorial‑grade content to researcher training and thought‑leadership initiatives, we help organizations engage scientific audiences in credible, meaningful ways. Explore how we can support your organization.