DCU Position Statement on the Use of Artificial Intelligence Tools

Actively engage with opportunities to develop or deepen your understanding of AI in your academic or professional area. All staff should take the DCU Introduction to AI Literacy course, designed by our internal team of academic experts in the field of AI, which supports our compliance with the new EU Artificial Intelligence Act. LinkedIn Learning, sector or disciplinary reports provide other rich sources of information.

Remain up to date on relevant guidance as it emerges and evolves nationally, internationally¹ and locally. Pay particular attention to those from research funders and publishers. DCU Guidelines such as those issued by the research Ethics Committee or the TEU  will be reviewed frequently and updated as needed to reflect changes in technology, regulations, and institutional priorities.

^{1 Examples include: Generative Artificial Intelligence: Guidelines for Educators, QQI 2023 and ERA Living Guidelines on the Responsible Use of AI in Research}

Keep in mind that you are fully responsible for your use or deployment of AI and responsible for acting with professionalism and accountable for any harm or misuse (even if unintended). Be transparent and document your use of AI. The use of generative AI for any University related research, scholarship or work, should be clearly disclosed to stakeholders and impacted parties or individuals.

Consider that generative AI-produced content may be biased, fabricated, inaccurate, or contain copyright-protected or proprietary information and will require human-review before use. It is not acceptable to use AI tools to create deep fakes or generate misleading content or to promote discrimination, bias, or harm.

Be alert to the particular risk with AI of inadvertently sharing data inappropriately or illegally. Your use of AI must be in compliance with existing University policies, which include:

• Data Privacy Policy (GDPR)

• Digital Systems and Cloud Services Policy

• Intellectual Property Policy

Ensure that you do not share any content that compromises an individual’s privacy, confidential university processes, DCU’s commercial, or strategic interests. If you do not know if it is safe to share DCU data with a specific AI tool, DO NOT. 

DCU data includes content from email accounts and drives, recordings or transcripts of DCU meetings, draft papers, books, book chapters, theses, or any document not officially published via our web pages or on DORAS. The use of Personal Data as an input to a GenAI tool is likely to constitute a data breach under GDPR, unless this action received explicit informed consent.

Data can be shared in a range of ways including:

• Information entered via prompts

• All uploaded material/data

• Personal identifying information such as names or email address

• Other background information you provide to improve personalization

AI product providers vary in their approaches to data protection, retention, and usage of your data. Entering information into AI tools is similar to posting that information on a public website. You have to assume that any information you input to, or give access to these tools is no longer private and may be used for training the model, sold on to 3rd parties, accessed by governments or made publicly available.  Open source European tools are a safer approach than using non-EU owned tools, which are very likely not GDPR compliant.

AI tools do not offer the 'right to forget' which means that once data is uploaded / entered - there is no way to delete it which can lead to loss of data ownership.

It is intended to build up a whitelist of tools that have been subject to scrutiny in terms of data privacy and licensing agreement conditions, in order to approve them for potential use with specific categories of DCU data. The approach will include a mechanism for colleagues to propose AI tools for review, and regular re-visiting of approvals with software upgrades or licensing changes. As with all IT investment, central licenses and quality controls will be important. Note that whitelisting will not supersede decisions to deploy an AI tool for a particular purpose, given that many applications of AI will involve important additional considerations.  

Under the University's existing enterprise licences, MS Co-pilot², Google Gemini AI and Zoom AI Companion suites are available to all staff. Login with your DCU identity. These systems are configured to be GDPR compliant, all data entered remains within the existing DCU data privacy agreements and no data is shared with any of these systems training models. No DCU data should be entered into any other AI tools.

^{2 This is distinct from Microsoft 365 Co-Pilot for which we have not entered into a licencing agreement.}

Protect your own intellectual property. Exposing information to an AI tool could result in lost ownership. 

Do not use AI detection software to review student’s work for plagiarism. Such tools are currently not particularly effective, but more importantly once submitted to an AI tool, the work can be harvested by the AI platform without student’s permission.