Digital Trust guidelines for Generative Artificial Intelligence use

Before using a generative AI tool, 

Intellectual property laws are designed for humans, and we’re just beginning to have legal cases to help establish how we’ll protect creative work made with AI tools. Generally, free generative AI tools assume that you are giving your prompts and inputs to them in return for the use of their tool. Use them with the assumption that all of what you give them and create with them is public.

 Reminder: “If you are not paying for the product, you are the product.”

Submitting Content to a Generative AI Tool

Generative AI tools learn through their interactions with users. For that reason, many generative AI tools require you to agree that they can keep the information you submit for future use, including by providing it to other users. This is especially true for free generative AI tools.

Do not share information that is confidential under university policies and rules.

Examples include:

• Student information
• Personnel information
• Information that the university has promised to keep confidential, for example, in a contract, a grant application, or a disclosure to research participants
• Information about the university that is confidential, proprietary, or otherwise not public

Do not share valuable intellectual property, whether it is yours, the university’s, or someone else’s.

The owner of intellectual property is entitled to control who can copy or use that intellectual property. By submitting information to a generative AI tool, you may give up valuable rights to control who can use that intellectual property in the future. This is especially problematic if you do not own the intellectual property or do not have the right to authorize others to copy or use it (for example, if the intellectual property belongs to ABOR, to another institution, or a co-author or collaborator). Be absolutely certain any information submitted to an AI-based model is public and does not include any sensitive or personally identifiable information - including information that could identify someone when combined with other data. 

ASU Policy Manual: 

ACD125: Computer, Internet, and Electronic Communications Information Policy 

ACD204-02: Code of Ethics

The output of a generative AI tool is a combination of the input you provide to the tool and the information the tool supplies from its data resources and processes. The more you interact with the tool, the greater your personal contribution to the resulting materials. For example, you choose what information to submit and what questions to ask, and you may also provide additional input, make follow-up inquiries, and edit, select, or compile the output of the generative AI tool to assemble the final product. 

Before using a generative AI tool to prepare materials, consider whether collaborating with the generative AI tool to prepare those materials will interfere with your ability to use the output in the way you plan. 

Be honest about your use of generative AI tools in preparing the materials. Do not hide or misrepresent your tool use and be prepared to answer questions about how the tool contributed to the materials. If the recipient of the materials is potentially unaware that you used a generative AI tool to create them, you should affirmatively disclose that you used a generative AI tool. For example, if you were asked to provide a writing sample as part of a job application, you should disclose to what extent a generative AI tool produced the end product, as the prospective employer would otherwise assume that you had prepared the materials yourself. 

Educate yourself about and follow any specific rules that apply to the preparation of specific materials, including any rules that restrict the use of generative AI tools in preparing information. This could include rules in a classroom syllabus or assignment or grant agreement provisions.

Consider whether you or the university need to own the intellectual property rights of selected materials. Materials prepared using a generative AI tool may not be eligible for intellectual property rights if there is insufficient direct human involvement in their development. Not owning intellectual property rights may impact your or the university’s ability to publish, distribute, patent, or use the materials. 

Various grant sponsors may take different approaches to using generative AI. For example, the NIH prohibits reviewers from using generative AI tools to analyze and critique NIH grant applications and R&D contract proposals, seeing them as a violation of confidentiality requirements. 

Understand that the international community may come to different conclusions on your rights regarding generative AI and intellectual property. The US recently launched an effort to understand the emerging space of generative AI and copyright. As of 2023, the United Kingdom has developed copyright and AI working groups. Europe is considering broader AI regulation, for which intellectual property will be a component. 

In the meantime, these questions may help you navigate ownership of intellectual property:

• Which humans are contributing to your work, and what is their contribution?
• What expectations does your team have regarding those contributions?
• What existing agreements or contracts might frame your work?
• Grant stipulations?
• Work for Hire agreements?
• Employee agreements?
• Where are your team members, and what international frameworks might your work fall under?
• Which AI tools are you using, and what are those tools contributing to your work?
• How do your human contributions significantly shape and define the final product?
• Do you have documentation of the design process and the inputs coming from your human teams?

By fostering a culture of awareness, responsibility, and compliance, we can ensure that the utilization of generative AI at ASU not only enhances our academic mission but does so in a manner that reflects our commitment to Principled Innovation. 

We recognize that the innovative application of generative AI presents a tremendous opportunity for enhancing learning, research, and administrative functions at Arizona State University. As we embark on this exciting journey, it is imperative that we adhere to a set of principles and guidelines that align with our institutional values, federal regulations and state laws. 

ASU’s Provost provides resources for generative AI in Teaching.

When you find a tool you would like to adopt, ASU has committed resources to help you assess and integrate technology into your work and classes. These processes will help you comply with university policy and state and federal laws, protect data privacy and security terms we have agreed to, and follow the risk mitigations we have put in place. 

Before you use a generative AI tool, you should understand whether you agree that the tool can keep or use the information you submit and, if so, whether that conflicts with your other responsibilities. 

Many of these tools are ‘click-through agreements’ and do not go through a formal procurement process. Using such tools as part of your role at the University may require taking on personal risk for using that tool. 

Leverage ASU’s Vendor IT Risk Assessment process to ensure that the tool has been properly reviewed. Additional supporting processes will help you should you need a tool to integrate into Canvas. 

For a complete list of AI tools already available at the university, please visit https://ai.asu.edu/ai-tools.