Artificial Intelligence at Augusta University

Artificial intelligence is reshaping the landscape of higher education, redefining the ways we teach, conduct research, and collaborate. Augusta University is dedicated to embracing these advancements with a focus on responsibility—fostering innovation while strictly safeguarding university data and upholding the highest standards of academic integrity. These institutional guidelines serve as a framework for our entire community, ensuring that students, faculty, and staff can explore the potential of AI within a secure and ethical environment.

Our core approach is built on five fundamental principles that ensure AI use aligns with our values of academic excellence and ethical conduct:

  • Protect University Data: Always classify your data before entering it into any system.
  • Prefer Approved Services: Use tools reviewed by AU IT for any non-public data.
  • Verify All AI Outputs: AI can "hallucinate" or create fabricated information; always review results before use.
  • Be Transparent: Disclose your use of AI whenever required by instructors or publishers.
  • You Remain Responsible: You are ultimately accountable for any work products or decisions produced with AI assistance.

Guidance by Audience

Students

#students

Faculty & Staff

#faculty-staff

Researchers

#researchers

Data Security and Tool Selection

Before you begin any project using artificial intelligence, you must identify the type of university data you intend to use. Augusta University classifies data into four categories to help you determine which tools are safe for your specific task.

Understanding Data Classifications

  • Public / Unrestricted: This includes information intended for everyone, such as our public website content or news. This data does not have restrictions on who can view it.
  • Sensitive: Internal information that isn't for public consumption and requires extra care to protect. While this data might be subject to open records requests, it still needs special security measures to prevent unauthorized access.
  • Confidential: Information subject to strict access restrictions to protect proprietary data or personal privacy.
  • Personal (PII): Data that identifies an individual, which must only be used for the specific purpose it was collected.

AI Tool Categories at AU

Selecting the right tool is just as important as identifying your data. Not all platforms offer the same level of security or contractual protection.

  • Public AI Tools (Unapproved): These are consumer-facing services like the free versions of ChatGPT, Claude, or Gemini. Because data protections may be unclear or subject to change, these may only be used with public, unrestricted information.
  • University-Approved AI Services: These tools, including Microsoft Copilot and Box, have been reviewed by AU IT and include strict security safeguards and data handling commitments.
  • Applications with AI Features: Many of our existing enterprise platforms, such as Zoom and Microsoft 365, now include built-in AI functionality designed to enhance your workflow.
  • Third-Party Add-ons and Plugins: These include browser extensions and meeting transcription bots. These represent the highest risk for data leakage and require a separate IT review before use.

University Decision Matrix

Data Type Public AI Tool (Unapproved) AU-Approved Service Third-Party Plugin/Add-on
Public / Unrestricted Allowed Allowed Use Caution
Sensitive Do Not Enter Only if approved for type Do not use unless reviewed
Confidential Do Not Enter Only if approved for type Do Not Use
Personal (PII) Do Not Enter Only if approved for type Do Not Use

Note that public tools like the free versions of ChatGPT or Gemini should only be used with public information. For any work involving sensitive university data, always utilize university-approved services like Microsoft Copilot.

Navigating AI Risks

While generative AI offers significant benefits, it also introduces specific risks that every member of the university community should understand. These risks range from how your data is handled to the legal implications of the content produced.

Data and Privacy Concerns

When you interact with an AI tool, the information you provide—including prompts and uploaded files—may be stored by the service provider for varying periods. Unless contractually prohibited, some services use this submitted content to train and improve their future AI models. Additionally, third-party plugins and connectors can transmit your data to outside parties, which significantly increases the risk of a security breach.

Accuracy and Reliability

It is critical to remember that AI can "hallucinate," meaning it can generate information that sounds perfectly plausible but is completely incorrect or fabricated. This extends to academic work, where AI has been known to create fake citations that appear legitimate. If you are using AI for technical tasks, be aware that generated code may contain bugs, security vulnerabilities, or logic errors.

Legal and Ethical Implications

AI outputs can sometimes resemble copyrighted material, leading to potential legal concerns. Within the university, the undisclosed use of AI in academic work may violate our academic integrity policies and be considered plagiarism. Furthermore, AI models can perpetuate societal biases or make unfair assumptions, which requires human oversight to correct.

 

Best Practices for the AU Community

Following these established best practices ensures that your use of AI remains secure, ethical, and productive.

Before You Start

  • Classify Your Data: Use the Augusta University framework to determine if your data is public, sensitive, confidential, or personal.
  • Check the Tool Category: Identify if the tool is a public service, an AU-approved application, or a third-party add-on.
  • Consult the Matrix: Verify that your intended use is permitted by checking the university decision matrix.
  • Ask IT: When in doubt, contact the Information Technology department before proceeding with any new tool.

During the Process

While working with AI, you should always minimize the amount of information you share and provide only the absolute minimum necessary for the task. Whenever possible, remove identifiers or aggregate data to protect individual privacy. You must never share secrets such as passwords, API keys, or other security-sensitive information. Most importantly, always verify AI-generated outputs against authoritative sources and keep records of AI use for any significant decisions.

After Using AI

Once you have generated your output, review it carefully for any potential bias or unfair assumptions. Be transparent by disclosing your use of AI as required by university policy or the specific context of your work. Ultimately, you remain responsible and accountable for all final outputs and decisions, regardless of how much AI assistance you received.

Student Hub: AI as a Learning Partner

For students, AI can be a powerful support system when used appropriately and transparently. It can help you brainstorm outlines, explain complex concepts in simpler terms, or help you edit and improve the grammar of your own original writing.

However, there are clear boundaries to maintain academic honesty. You must always follow the specific AI guidance provided in your course syllabus. The following table outlines the general "Do's and Don'ts" for AU students:

Acceptable Academic Uses (When Permitted) Prohibited Academic Uses
Brainstorming and outlining Submitting AI-generated work as original work
Creating study materials and practice problems Using AI during exams unless explicitly allowed
Language improvement on your original work Circumventing core learning objectives

Key Principles

  • Follow your syllabus: Always check course-specific AI policies before using AI tools.
  • Never claim AI work as your own: Submitting AI-generated content as original work when not permitted is academic dishonesty.
  • Disclose your use: When AI use is allowed, cite it appropriately as required by your instructor.
  • Understand the learning goals: Consider whether AI use supports or undermines your educational development.

Faculty and Staff: Professional & Academic Leadership

As leaders in our academic community, faculty and staff play a critical role in modeling the responsible use of AI. Whether managing a classroom or a departmental meeting, the goal is to balance the efficiency of these tools with our commitment to security and academic integrity.

Guidance for Faculty and Instructors

Instructors should establish clear AI policies for their courses and communicate them explicitly in both the syllabus and individual assignment instructions. A proactive approach helps students understand the boundaries of acceptable use while fostering AI literacy. We recommend the following strategies for your curriculum:

  • Specify Allowed Uses: Clearly define what level of AI assistance is permitted for each specific assignment, as "acceptable use" may vary depending on the learning objectives.
  • Require Disclosure: Ask students to document and cite their AI use when applicable to ensure transparency and proper attribution.
  • Design AI-Resistant Assessments: Consider developing assignments that emphasize the creative process, personal reflection, or direct application of knowledge, which are more difficult for AI to replicate effectively.
  • Teach AI Literacy: Help students navigate the evolving landscape by discussing both the capabilities and the significant limitations—such as inaccuracy and bias—of current AI tools.

AI in Meetings and Daily Operations

For staff and faculty hosting or participating in university meetings, maintaining data privacy is a primary responsibility. To protect sensitive discussions, please adhere to the following protocols:

  • Manage Access: Use meeting waiting rooms and require university authentication for any sessions involving non-public information.
  • Transparency in Recording: Always announce at the start of a meeting if AI transcription or recording features are active.
  • Protect Sensitive Data: Disable AI transcription or recording features immediately if the conversation shifts toward sensitive, confidential, or personal information.
  • Monitor for Unauthorized Bots: Be vigilant for unauthorized third-party transcription bots and report them to the meeting host and IT immediately.

By setting these clear expectations, we can ensure that AI serves as a tool for enhancement rather than a risk to our institutional standards.

AI in Research: Innovation with Compliance

While AI offers powerful capabilities for data synthesis and analysis, researchers at Augusta University must exercise extreme caution when handling unpublished, proprietary, or regulated data. Because the integrity of your findings and the security of your intellectual property are paramount, all AI use in a research context must align with university-approved protocols and external sponsor requirements.

Critical Considerations for Researchers

  • Protect Unpublished Findings: You must never enter unpublished research data or proprietary findings into unapproved or public AI tools.
  • Review Grant and Sponsor Agreements: Specific contracts may have clauses that strictly restrict data sharing, AI usage, or the disclosure of intellectual property.
  • Human Subjects and IRB Approval: Any research involving human subjects requires formal IRB approval before AI can be used to process or analyze participant data.
  • PHI and HIPAA Compliance: If your work involves Protected Health Information, you are restricted to using only specially approved, HIPAA-compliant AI tools.
  • Maintain Reproducibility: To ensure the scientific integrity of your work, you must document your AI use—including specific prompts and model versions—whenever AI influences your methods or results.
  • Authorship and Attribution: Always follow the specific requirements of your discipline or target journal regarding the disclosure of AI assistance in your manuscripts.

Acceptable Research Uses

When using approved datasets and following university privacy guidelines, researchers can utilize AI for several high-value tasks:

Literature Review

Synthesizing published works to identify trends or gaps in existing research.

Data & Code

Assisting with code generation for data processing or performing analysis on approved, de-identified datasets.

Writing & Grants

Editing manuscripts for language improvement and supporting grant writing for non-confidential sections.tex

Reporting Incidents and Getting Help

Early reporting is critical to minimizing harm if sensitive data is accidentally shared. If you realize you have entered confidential or restricted information into an unapproved tool, follow these steps:

  1. Stop immediately and do not enter any more data.
  2. Document the incident, noting what occurred and which tool was used.
  3. Report to Cyber Defense by emailing 72cyber@augusta.edu or calling (706) 721-9237.

For general questions about software approval or technical setup, contact the IT Help Desk at (706) 721-4000.