Guiding Principles on the Responsible Use of Generative AI in Research

(Approved by Senate Research Committee on 24 September 2026; noted by Senate on 20 October 2026)

• PDF

Recent advances in artificial intelligence have substantially enhanced research and development activities. In particular, generative AI (GenAI) has emerged as a valuable tool that enables researchers to accelerate their work and broaden the scope of inquiry. HKUST encourages the responsible use of GenAI as a research augmentation tool provided that human oversight, accountability, research integrity, and transparency are maintained throughout the research process. While recognizing the opportunities GenAI offers, the University also acknowledges the potential risks associated with its use in research. This document intends to state the overarching guiding principles researchers need to abide by to ensure responsible use of GenAI in research ¹. These guiding principles align with the University’s Policy on Research Conduct and Integrity (Policy), and any practice that runs counter to these principles will be reviewed in accordance with the Policy.

Authorship

HKUST members are required to uphold the highest standards of research integrity to ensure that all research remains accurate, trustworthy, and ethically sound. GenAI is not considered an author of a publication and does not hold ownership of the research outcome. HKUST members remain ultimately responsible for their use of GenAI-generated content, including ensuring its integrity, accuracy, and originality. Regardless of the tools used, it remains the researcher’s responsibility to verify that all research processes and content involving GenAI comply with applicable laws, regulations, institutional policies, specific policies of funding agencies or sponsors, and editorial policy of specific journals or publishers, and do not constitute research misconduct.

Editing

GenAI may be used to support language editing and translation, provided that such use does not replace the author’s intellectual contribution, critical judgment, or scholarly responsibility. All AI-assisted content must be reviewed and verified by the author to ensure that the final work accurately reflects the researcher’s expertise and original contribution.

Accuracy

AI tools must not be used to fabricate, falsify, misrepresent research data, findings or analyses, or distort scholarly content. Researchers should recognize the risk of data fabrication, falsification or AI hallucination and take steps to mitigate it. Authors are required to fact-check all AI-generated materials for accuracy, neutrality, and appropriateness, and to take appropriate steps to identify and mitigate bias, stereotyping, or misinformation in AI outputs. This includes verifying that all claims, citations, references, and analyses are correct, complete, and consistent with the authors’ own scholarly judgment. AI tools may be used to identify potential hallucinations and plagiarism, including fabricated references and unsupported or uncited claims. The use of AI does not diminish the author’s accountability for the final content of the work.

Disclosure

The use of GenAI to generate or substantially contribute to research content, including text, images, audio, video, software code, synthetic data, analyses, or other materials incorporated into research outputs, should be transparently disclosed and documented.

Oversight

When GenAI is used in intellectual interpretation or scientific judgement, such as using GenAI in hypothesis, data analysis, data interpretation, etc., the researcher needs to exercise substantive human oversight and retain ultimate responsibility for validation, interpretation and conclusions. When conducting peer review, GenAI must not compromise confidentiality or substitute for the reviewer’s judgment. Where GenAI is used to assist intellectual judgement, its role should be clearly disclosed, the key arguments, conclusions, and other content being influenced should be clearly identified and documented, with an explanation of how the researchers reviewed and verified all AI‑generated content.

To comply with the policies of funding agencies, publishers, and collaborative partners, the use of GenAI in material should be clearly disclosed. It is recommended that researchers maintain a clear record of all AI technologies employed, including the specific tool and version, as well as the date of content creation.

Confidentiality

Researchers must protect confidential, proprietary, personal, and otherwise restricted information at all stages of research, and must ensure that transmitting any such data to a GenAI model is legally permissible. Sensitive or restricted data must not be entered into external AI systems unless explicit authorization has been granted and appropriate safeguards are in place. Researchers should also recognize that AI services may retain user inputs or generate outputs that contain personal data, and that such information must be handled in accordance with applicable privacy and data protection requirements.

Terms of use

Before using any AI technology, researchers must understand the applicable terms and conditions, terms of use, and any licenses or contractual restrictions associated with the tool. Researchers must confirm that the GenAI tool does not claim ownership of their content or impose limitations that would interfere with the researcher’s rights to use specific outputs. Researchers are responsible for ensuring that their use of AI does not violate copyright, licensing, or other intellectual property or confidentiality obligations.