AI Guidance

AI Guidance

AI creates opportunities for teaching, research, and administrative efficiency, but it also introduces significant risks that can affect students, faculty, staff, research programs, and institutional compliance so it must be used thoughtfully and responsibly.

This guidance outlines important practices for protecting sensitive information, evaluating AI-generated content, maintaining human oversight, and complying with UTA policies and State requirements. By following these principles, employees can take advantage of AI’s benefits while helping protect university information, maintain public trust, and remain accountable for the work and decisions AI supports.

  • Before entering information into an AI tool, consider whether it contains personal information, confidential records, protected data, or other sensitive State information. Privacy and security protection must always be maintained.
  • AI systems can introduce cybersecurity risks, including improper disclosure of information, malicious prompts, and vulnerabilities. Users should use only approved AI tools and follow security requirements. AI tools must be submitted for technology assessment.
  • AI should only be used for approved business purposes and within established acceptable-use policies. Users should understand the tool’s intended use and limitations.
  • As the Information Resource Owner, you are accountable for the information you provide to AI tools and for how you use their outputs
  • Bottom Line – Know your Data!

  • AI can generate inaccurate, incomplete, or fabricated information. Users must review and verify outputs before relying on them.
  • As State employees, you are responsible for accuracy of AI outcomes.
  • AI systems can create unfair or discriminatory outcomes if data is incomplete, inaccurate, or unrepresentative. UTA employees should carefully evaluate results that affect individuals or groups.
AI should support human decision-making, not replace it. Humans remain responsible for reviewing outputs and making final decisions, especially when actions affect people, services, or public interests.
  • Users should know whether prompts, inputs, or outputs are stored, shared externally, or used to train models. Texas AI governance materials require attention to data sharing, retention, and security. Understand that your data may be subject to records management requirements, retention schedules, audits, or public information requests.
  • Texas law now requires AI training for UTA employees. Effective AI use requires understanding AI capabilities, limitations, risks, ethics, and best practice.
  • When AI is used to create content, support decisions, or interact with the public, users should ensure any required disclosures are made and that stakeholders understand AI’s role. Transparency is a core requirement of Texas AI governance.

Some AI service tiers may allow user prompts, uploaded files, or interaction data to be used for product improvement or model training unless data-sharing settings are turned off. While the exact steps may differ across AI tools and account types, the core concept is the same: review the privacy or data control settings and disable any option that allows your data to be shared or used to improve the model. Here are some examples.


AI Tool Model-Training Setting to Turn Off
ChatGPT Improve the model for everyone
Google Gemini Gemini Apps Activity / Keep Activity
Claude Help improve Claude

Steps for each AI tool

ChatGPT Google Gemini Claude
1. Open ChatGPT 1. Open Gemini 1. Open Claude
2. Select your profile picture 2. Select your profile picture 2. Select your profile icon
3. Open Settings 3. Open Gemini Apps Activity 3. Open Settings
4. Go to Data Controls 4. Turn Gemini Apps Activity off 4. Open the Privacy section
5. Turn Improve the model for everyone off 5. Optionally, delete existing stored activity 5. Turn Help improve Claude or a similar model-improvement setting off

A Simple Rule of Thumb

Before using AI for State business know the “Ws”, ask yourself:

Can I explain:

  • What information I entered?
  • Was the information appropriate to share?
  • How I verified the output?
  • Who is accountable for the final decision?
  • Did the AI use comply with UTA policy and Texas AI ethics requirements?

If you cannot confidently answer these questions, pause and seek guidance before relying on the AI output.