No. Uploading entire copyrighted materials, such as journal articles or book chapters, to AI tools is generally prohibited by most publishers’ licensing agreements.
These agreements typically restrict sharing full-text content with third-party platforms (including AI tools) to protect intellectual property rights. While some AI tools claim that uploaded content remains private, always review the specific terms of both the content provider and the AI platform before proceeding.
Instead, consider using AI to summarize your own notes.
For copyrighted content, limit input to very brief excerpts and ensure proper citation.
Generally, yes. Open access materials are often licensed under Creative Commons (CC), which typically allows the use of works with AI tools.
However, different CC licenses have different requirements, e.g. CC BY-NC (non-commercial) restricts usage to non-commercial purposes. Always check the specific CC license attached to the material you intend to use and ensure your use complies with its conditions.
Training AI models with large text and image datasets often involves text and data mining (TDM), which is subject to specific licensing agreements.
Many publishers support TDM for research purposes, but may have restrictions, such as download limits or require certain fees. Check our guide for different TDM policies of publishers and aggregators.
Open access materials under Creative Commons (CC) licenses or in the public domain generally do not require additional permissions.
However, some OA content may include third-party materials (e.g. images or figures) that are subject to stricter copyright terms, requiring you to seek additional permissions. Check this article to learn more about using CC-licensed content in AI training.
If your dataset contains personal data, you must ensure compliance with privacy regulations. Consult with your supervisor or senior researchers on any established protocols for handling such sensitive data.
University’s and the Government’s policies on data privacy can also help:
Copyright owners can take additional steps to prevent AI tools from using their work, such as:
Check policies on AI training via the Generative AI Licensing Agreement Tracker.
How Copyright Relates to GenAi from University College of London (UCL): while HKSAR law and British law are different, the HKSAR takes reference from UK law and other common law jurisdictions, and the language of the ordinance for computer generated work is quite similar.
The HKSAR is a separate legal jurisdiction from the rest of China. However, it is useful to be aware of some recent issues in copyright and GenAI in China.
