AI Policy Menu

Policy on Generative Artificial Intelligence

Who Is This Policy for?

This policy applies to any author who submits a research proposal, report, article, award application, or other work to the Natural Hazards Center. This includes applicants to our Research Award Programs, awardees, and other authors or contributors.

Our Position and Background on This Policy

At the Natural Hazards Center, we respect and embrace technological change; we also bring a healthy sense of caution as related to the transformative power and potential dangers of artificial intelligence (AI). Because we are committed to scholarly discovery as well as applying knowledge to reduce the harm caused by disasters, we are especially attuned to the social, educational, and environmental risks of AI.

We developed this policy regarding generative AI tools—specifically those based on Large Language models (LLM) and used to generate text, imagery, audio, and synthetic data, such as ChatGPT, Claude, Google Gemini, Microsoft Copilot, NovelAI, Jasper AI, Rytr AI, DALL-E, and others—because they are dramatically impacting the research enterprise. As we learn more about the evolving effects of AI on research and its applications, we will update this policy.

Does the Natural Hazards Center Allow Use of Generative AI?

Authors must develop and present their original ideas in all work they submit to the Natural Hazards Center. As a matter of scholarly integrity, generative AI tools should never be used to create data, original scholarly writing, or other related outputs.

We understand that researchers are increasingly using AI tools for many purposes, which may be appropriate in some limited or specific circumstances. Using AI, however, comes with significant risks—including violations of confidentiality, plagiarism, fabricated citations, introduction of bias, erosion of trust in science, and other forms of academic misconduct.

The bottom line is that your professional work should be your own and any use of AI tools that undermines academic integrity will be considered academic misconduct. The Center will not review proposals or other submissions that fail to comply with the guiding principles outlined in this policy. Authors who violate these principles will be deemed ineligible for funding or recognition through the Center’s various research award and scholarship programs, and we will discontinue support for future publications or presentations.

Guiding Principles

To maintain integrity and minimize some risks posed by AI, the Natural Hazards Center has adopted the four guiding principles outlined below. For context, the Center follows the Publication Manual of the American Psychological Association (APA, 7th edition) style guide and reporting standards. The APA recently adopted its APA Journals policy on generative AI, and we have adapted their policy into these principles.

  1. Maintain Ethical Rigor. AI tools should never be used in ways that compromise the confidentiality of your research participants, co-authors, or other collaborators. Steps to maintain an ethical approach to AI use are listed below. This is not a comprehensive list.

    • Protect confidentiality. Authors should be aware that it is a violation of confidentiality to share data that contains personal identifying information or other confidential disclosures with AI companies. Therefore, AI tools—including transcription or translation services—should only be used when all confidential information has been removed.

    • Obtain approval from research participants, collaborators, and co-authors. Research participants, collaborators, and co-authors may object to having their original ideas and personal thoughts and stories shared with AI. To respect their right to privacy, obtain their approval before using AI tools.

    • Use the consent process to obtain informed consent from research participants. If your research involves human subjects, authors should disclose any planned use of AI tools and their associated risks to research participants during consent procedures.

    • Obtain institutional approval. Planned AI use must also be disclosed in Institutional Review Board (IRB) applications and funding proposals. You should ensure such use receives IRB and funder approval.

  2. Take Responsibility. AI should never be used without human oversight and critical judgment. Authors must take responsibility for all AI-generated output. This requires reviewing and evaluating the output, verifying its accuracy, and mitigating AI-related bias. Examples of the ways you must take responsibility for AI-generated output are listed below. This is not a comprehensive list.

    • Literature reviews or gaps in knowledge. If you used AI to summarize past research or identify gaps in knowledge, you must read the original articles AI cites as sources. AI is known to mischaracterize source material, fabricate sources that do not exist, cite research that does not support its claims, and generate biased claims or arguments.

    • Research instruments. Researchers must critically assess any AI-generated research instruments (e.g., survey instrument, interview guide, visual prompt). A trained expert in the research method should confirm the instrument’s validity and reliability.

    • Language translation. If AI is used to translate a research instrument, interview, or other text into another language, authors should have the output checked by someone who speaks the language.

    • Transcription. Authors should verify any audio or video files transcribed by AI by listening to the audio or video file while reading the AI transcription and correcting any errors.

    • Code or analysis. Authors must review and verify any AI-generated statistical code or AI-assisted quantitative or qualitative analysis.

    • Tables or figures. Authors must verify that any tables or figures generated by AI accurately represent the data or other content they claim to graphically depict. Authors must disclose how they used AI to generate the table or figure in the note below the item.

    • Reference lists. To address the proliferation of AI-hallucinated sources in reference lists, authors must verify their reference entries are accurate and enable readers to access the original source material by including Digital Object Identifiers (DOIs) or permanent web addresses (URLs), when these are available.

    • Photos or videos. AI tools should not be used to generate or alter photos, videos, or audio of real people, places, or events. Authors must not submit such items, unless their use was approved as a research instrument by their IRB. Please contact the Center for further review. Authors must also avoid using any AI-generated images of real people, places, or events which have been produced by third party vendors (e.g., Shutterstock) or other sources.

  3. Be Transparent. Authors who used generative AI must disclose it. This includes any use (or planned use) of AI to create research instruments, collect or analyze data, generate or refine code, transcribe audio or video files, translate text, or any other research element. It also includes using AI for brainstorming, summarizing past research, generating or revising written content, creating tables or figures, editing, proofreading, translating, or any other phase of the writing process.

  4. Provide Attribution. Any AI use should be disclosed and cited in the body of the proposal or paper and in an author note. If you used AI in multiple ways, each instance should be disclosed and cited. To learn more about the Center's AI disclosure requirements, including how to cite AI chats and tools in the body of your submission and prepare an author note, click on the link below.

Read the Natural Hazards Center's Instructions for Disclosing AI Use here.

If you are submitting a proposal, report, or other written work, you will be asked to disclose any AI use as part of the submission process. Please follow the directions you are provided closely.

Our Commitment

When a member of the Natural Hazards Center team or one of our partners is reviewing a proposal, report, article, or other submission to the Center, we always treat the document as confidential. Members of our review team will not upload the manuscript or any part of it into a generative AI tool as this may violate the authors’ confidentiality and proprietary rights, and, where the paper contains personally identifiable information, may breach data privacy rights. Further, we are committed to always developing original reviews and assessments with our own words and writing.

Further Reading

Other research organizations, agencies, and publishers have developed their own AI use and disclosure policies. We have provided links to some of these resources that have helped inform our own processes. We encourage others to review them as well.

American Psychological Association (APA)

Federal Agencies

Other Resources

Questions?

If you have questions or suggestions regarding this policy, please reach out to us at hazctr@colorado.edu.

If you would like to reference this policy, see the suggested citation below:

Natural Hazards Center. 2026. Policy on Generative Artificial Intelligence. https://hazards.colorado.edu/research/nhc-policy-on-use-of-generative-ai

Published: May 2026