Generative AI Usage Guidelines

These usage guidelines are intended to help you use Generative Artificial Intelligence (GenAI) tools safely and sensibly. In addition, these guidelines provide guidance on what GenAI can and cannot be used for.

There are many definitions of Generative AI. The definition that we at EUR use is the definition that the central government also uses:

Generative AI is a form of AI in which complex algorithms are used to generate new content such as text, images, computer code or videos.

Artificial Intelligence (AI) is a collective name for algorithms and methods that perform tasks that were thought to require human intelligence. Artificial intelligence refers to systems that exhibit intelligent behavior by analysing their environment and - with a certain degree of independence - taking action to achieve specific goals. It is not just about computing power, but about the ability to learn (independently) and make decisions. The learning ability is therefore typical of artificial intelligence. AI uses rules formulated by humans or composed by the algorithm based on the data and trains itself with data.

These guidelines are published on MyEUR as a basis for future AI policy within EUR and for references to other sources.

Introduction

With the emergence of ChatGPT, Copilot, Gemini and other LLM (large language model) generative AI (artificial intelligence) tools, hereinafter collectively referred to as 'GenAI tools', many members of the EUR community are excited to explore their use in the university context. This advisory provides guidance on how to use these GenAI tools safely, without compromising institutional, personal, or proprietary information. Additional guidance may follow as conditions develop. In addition, this advice provides guidelines for the use of GenAI tools and refers to internal and external relevant sources for more information.

We choose to use GenAI responsibly. It is important that staff, faculty, and students learn how GenAI tools work so that they can maintain academic standards and maintain ownership of the written text.

If you have any questions about and/or additions to these guidelines, please send a message to Documentary Information Management (CIO Office) at DIMAdvies@eur.nl.

The golden rules for using GenAI

GenAI is a tool

• Use GenAI as a sparring partner, for example to brainstorm about topics. Please note that the information may be incorrect.
• Use GenAI as a supplement and not as a replacement. Stay critical.
• Use GenAI to boost your creativity. It can be a good way to discover new perspectives.
• Use GenAI to create summaries.
• Use GenAI to create a basic text based on your own input.
• Use GenAI to make your own text accessible to a specific target group.

Always check and edit the text before sharing it with others. Ultimately, you remain responsible for the work you deliver and share.

GenAI is boring

Always adapt GenAI's output to your own writing style and that of the EUR before using it. Also view the EUR Style Guide. Of course you want to prevent the reader from seeing that your text has been generated by a GenAI tool and resembles texts from others.

Use your common sense

Would you also ask a student or colleague the same question you ask a GenAI tool? If not, then you don't need to ask a tool.

GenAI is an editor

Use GenAI to set up or structure texts or for inspiration when writing your own work. Using whole sections of texts from, for example, ChatGTP or Copilot literally is not only boring, but can also lead to plagiarism.

GenAI is not a search engine

Do not use GenAI for fact-finding or fact-checking. GenAI tools give you a nice answer to your question, but there is no guarantee that this answer is correct and reliable.

GenAI does not care about Privacy

Do not mention personal information about yourself, others or the EUR in a question to GenAI. You should not share personal or confidential information with GenAI tools. There is currently no EUR-wide contract that specifically protects such data for reuse by the GenAI model.

In general, personal data covered by the GDPR, such as health information, as well as business information and any other information classified as confidential or controlled university data, may not be used with GenAI tools.

Everything you enter into a GenAI tool is used to improve its own model, which then becomes public data for everyone who uses the tool. This could lead to a data breach. If you suspect that you have shared confidential information, report this as a possible data breach to the IT Service Desk. A timely report ensures that the risk can be limited.

Data that is publicly available or defined as published university information ('Public' according to EUR Data Classification Policy) can be freely used in GenAI tools. In all cases, use must be in accordance with the Acceptable Use Policy.

Learn to use GenAI

If you ask a question in a GenAI tool, include background information or examples yourself so that the tool can respond accordingly. We recommend that you learn how to 'prompt' by following a training course and using good examples. Searching the internet will yield many paid and unpaid training courses. Training courses on the use of GenAI are also regularly given at EUR.

Always check the reliability of the information you receive via GenAI. Make sure you verify everything with other sources to check reliability. 

Consider ethical aspects, including respect for intellectual property and privacy. Also check the generated text for inappropriate content.

Use GenAI lawfully

Please note that the company that owns a GenAI system, for example ChatGPT or OpenAI, explicitly prohibits the use of the tool and their other products for certain categories of activities, including fraud and illegal activities. This list can be found in their usage policy. AI tools of any kind may not be used for activities that are illegal, fraudulent or in violation of any law or EUR policy.

Are you unsure about the appropriate use of GenAI? As a student, please contact your training coordinator or as an employee, contact your Information Manager.

GenAI in education

• The basic principle is that you always create and submit your own work. You bear full responsibility for your work. If it is not your own work, or generated by a GenAI tool, then that is fraud.
   
• If you have used GenAI somewhere, be transparent about it. Make sure it is clear how you have used generative AI and how you have ensured its responsible use.
   
• Any form of literal copying and copying without full source acknowledgment (quoting, referencing) of material generated by GenAI is not permitted without the permission of the teacher. After all, academic work requires checking the original sources. 
   
• Any use of GenAI during exams on-campus or other evaluations where it is indicated that the use of Generative AI is not permitted is fraud and is not permitted.
   
• Ask your own teacher, department or faculty whether GenAI is or is not allowed in your education and research and how you can use GenAI.

For more information, see this page of the Community for Learning & Innovation: Generative AI in education. 

What should you not use GenAI for?

GenAI tools may not be used for legal or assessment issues without explicit permission. 

Examples of this are:

• Generating proprietary or unpublished research
• Legal analysis or advice
• Recruitment, personnel or disciplinary decision-making
• Assessment

This list is far from complete. Please be aware that GenAI is and remains a tool and not an authorized source or decision-making tool.

Using your EUR email address

You are only allowed to use an EUR e-mail address to create an account for applications you need for your work.
According to articles 2.2 and 5.1 of the ICT Regulations, it is forbidden to use the EUR e-mail address for private accounts. The EUR e-mail address should only be used for work-related applications, unless explicit permission has been granted.

Commercial Data Protection in Copilot

When using the web version of Microsoft Copilot (https://copilot.microsoft.com), Commercial Data Protection is enabled.

Commercial Data Protection ensures that a lot of data is not kept permanently by Microsoft and that user and organisational data is protected. Below you can find what data is and is not collected and how long it is kept.

What is collected:

• Prompts and outputs: The questions you ask Copilot and the answers you receive. This data is kept for a short period of time (during the session).
• User data: This is your username, email address and the names of documents you upload. This data is kept permanently.
• Metadata: Information such as timestamps and technical details about how you use Copilot. This data is kept temporarily for analysis and improvement of the service.

What is not collected:

• Contents of private files: The specific content of your private files is not stored by Microsoft or used for other purposes.
• Personal conversations: Private conversations and personal communications are not collected or analysed.

Data held temporarily and permanently are not guaranteed to remain within EU borders. Users who have access to Erudite are strongly advised to use Erudite as a tool.

Please note that if you are participating in the M365 Copilot trial, this guideline does not apply to you. The standard ICT guidelines and terms of use apply to this small group of users.

More information

There is a lot of information available within EUR, SURF and other universities. Some references are:

• Regulations governing use of internet and IT facilities (2021)
• SURF: Artificial Intelligence | SURF Communities
• From education: AI voor Docenten | Uw bron voor AI in het onderwijs (in Dutch)
• University Leuven: Responsible use of Generative Artificial Intelligence 
• For further guidance on using ChatGPT and other AI tools for education and learning, see the Community for Learning & Innovation (CLI) guidelines.

Why these guidelines?

All content entered into or generated by GenAI is available to the AI ​​Model owner and their employees. There is currently no agreement between EUR and an AI tool supplier that would provide data security and privacy protection as required by EUR and Dutch law (GDPR). Consequently, the use of ChatGPT/Copilot/Gemini or other AI tools at this time may expose individual users and EUR to potential loss and/or misuse of sensitive data and information.