More and more Goizueta students are taking advantage of a generative AI’s ability to summarize and organize course content. This application is becoming a commonly-used tool for streamlining heavy course readings and leveraging efficiencies for team collaboration and research synthesis.
Makes sense! But did you know that you are likely infringing copyright?
If you are doing any of the following, you are placing yourself and possibly your university/organization at risk:
- Uploading any market research reports, articles, spreadsheets, images etc. from any databases licensed by Emory Libraries, your workplace, etc.
- Uploading any content accessed from Google without validating your copyright permissions.
- Uploading your faculty syllabus, project descriptions, etc.
- Easy solution. Ask your faculty for permission.
- Uploading IMPACT team research, deliverables or client-provided documents; applies to all NDAs.
Wondering Why This Applies to Google?
- Copyright infringement: Most content found on Google, including images, text, and videos, is protected by copyright. Uploading and using copyrighted material in a generative AI can be considered a form of infringement.
- “Fair use” defense: While generative AI companies argue that using copyrighted content for training falls under “fair use,” this is an unsettled legal argument.
- Personal and sensitive data: Google’s policies strictly forbid inputting personal or sensitive information into generative AI. Never presume the generative AI is a private space, as any data you enter can be collected, stored, and potentially used by the AI company.
- Terms of service violations: Using content from Google’s various services in a way that violates their specific terms of service is prohibited. Google’s terms explicitly state that you may not use its services or content in fraudulent or deceptive ways, or to violate others’ intellectual property rights.
Practical Risks of Using Content from Google
- Legal liability: You, not the AI, are responsible for the content you generate. If your output infringes on someone else’s copyright, you could be held liable for the damages. Claiming that the AI created the content does not absolve you of this responsibility.
- Misleading output: The AI model might “regurgitate” portions of the copyrighted content it was trained on, or produce content that is “substantially similar” to the original. If this happens and you use the output, you could be liable for copyright infringement.
- Unprotectable AI output: Even if you use an AI to create new work based on an original Google image, the resulting output may not be protected by copyright. The U.S. Copyright Office has stated that work generated purely by an AI without sufficient creative human input is not eligible for protection.
Questions?
Email GBSaskalibrarian [at] emory [dot] edu
Sources
Martenson LLP, AI copyright Infringement: Legal Riska & What to Know
University of South Florida Libraries, Copyright and generative AI.