Responsible Use
AI tools can help with drafting, summarizing, brainstorming, research preparation, file analysis, and coding. You remain responsible for what you enter, what you generate, and how the output is used.
AI output should be reviewed before it is submitted, published, sent, filed, used in a decision, or treated as evidence.
Quick Links
ISO AI Services Decision Matrix: https://security.utexas.edu/iso-policies/cloud-services/decision-matrix#ai-services
UT Acceptable Use Policy: https://security.utexas.edu/policies/aup
Acceptable Use of Generative AI Tools: https://security.utexas.edu/ai-tools
UT Austin information security policies: https://security.utexas.edu/policies
Responsible adoption of AI tools: https://provost.utexas.edu/the-office/academic-affairs/office-of-academic-technology/responsible-adoption-of-ai-tools/
OpenAI Usage Policies: https://openai.com/policies/usage-policies/
Anthropic Useage Policy: https://www.anthropic.com/legal/aup
Official Policy, Usage, And Data Links
Provider | Official page | What it covers |
|---|---|---|
UT Austin | https://security.utexas.edu/iso-policies/cloud-services/decision-matrix#ai-services | AI service data-use guidance |
UT Austin | UT guidance for generative AI tools | |
OpenAI | Acceptable and prohibited uses for OpenAI services | |
Anthropic | Acceptable and prohibited uses for Anthropic services |
Do Not Paste Without Approval
Do not enter sensitive, confidential, protected, unpublished, or personally identifiable information into ChatGPT or Claude unless the use has been approved.
Use extra care with:
Student records or education records.
Grades, advising notes, accommodations, or conduct information.
Personal information.
Health, financial, legal, or employment information.
Confidential University business.
Unpublished research.
Proprietary, copyrighted, or licensed material.
Passwords, credentials, access tokens, or security details.
Source code or logs that contain secrets or private data.
What Data Can I Use With Claude and ChatGPT?
The tables below summarizes the Claude and ChatGPT rows from the ISO AI Services Decision Matrix.
Claude service | Published Data | Controlled Data | HIPAA | FERPA | SSNs | PCI | Export controlled | IRB | CUI |
|---|---|---|---|---|---|---|---|---|---|
Anthropic Claude for Education | Yes | Yes | No | Yes | No | No | No | No | No |
Anthropic Claude for Education HIPAA | Yes | Yes | Yes | Yes | No | No | No | No | No |
Anthropic Claude API | Yes | Yes | Yes | Yes | No | No | No | No | No |
Anthropic Claude CoWork/Code | Yes | Yes | No | Yes | No | No | No | No | No |
ChatGPT Service | Published data | Controlled data | HIPAA | FERPA | SSNs | PCI | Export controlled | IRB | CUI |
|---|---|---|---|---|---|---|---|---|---|
OpenAI ChatGPT Education | Yes | Yes | No | Yes | No | No | No | No | No |
OpenAI ChatGPT Education HIPAA | Yes | Yes | Yes | Yes | No | No | No | No | No |
OpenAI API | Yes | Yes | Yes | Yes | No | No | No | No | No |
Review Before Using Output
Review area | What to check |
|---|---|
Accuracy | Are the facts, numbers, names, dates, and claims correct? |
Sources | Do citations, links, and quotes exist and support the claim? |
Context | Does the answer fit UT policy, course expectations, unit procedures, and audience needs? |
Bias and tone | Could the output be unfair, exclusionary, misleading, or inappropriate? |
Privacy | Does the output reveal information that should not be shared? |
Authority | Does a human expert, supervisor, instructor, or owner need to review it? |
When To Pause
Pause before using AI when:
The task affects a student, employee, patient, applicant, or member of the public.
The output could influence a high-stakes decision.
The content involves legal, medical, financial, safety, or employment matters.
The source material is confidential, sensitive, regulated, or unpublished.
You are unsure whether the tool or account is approved for the data.
FAQ
Does vendor access mean the data is approved?
No. Access to a tool and approval to use a specific data type are separate questions.
Can I paste student records?
Check UT guidance and the ISO matrix before using AI with student records or education records.
Can AI output be used as final work?
AI output should be reviewed by the responsible human before it is used, submitted, sent, or published.
Support
If you need help with... | Use this contact or page |
|---|---|
Data classification, IRB, HIPAA, FERPA, or sensitive data | |
General AI Studio questions | |
Course or assignment expectations | Course instructor or course staff |
Department workflow questions | Unit owner, supervisor, or process owner |