DATA-04 Artificial Intelligence Policy

Table of Contents

When using AI...
Never provide information you would not be comfortable posting on a public website unless the tool has been specifically approved to protect that information.

Three Major AI Use Cases

Artificial Intelligence (AI) can be used in different ways ranging from simple conversations to fully autonomous systems. As AI capabilities increase, so does the need for oversight, security controls, and governance.

This policy framework recognizes three common categories of AI use:

1. Conversational AI (AI Chat)

The most common use of AI is conversational assistance. Users interact directly with an AI system through prompts and receive responses in the form of text, images, code, summaries, recommendations, or other content. Typical uses include:

  • Brainstorming ideas
  • Drafting emails and documents
  • Summarizing articles and reports
  • Creating outlines and lesson plans
  • Generating or debugging code
  • Learning assistance and tutoring
  • Research support and information gathering

In this model, the human remains responsible for all decisions and actions. The AI provides suggestions, but it cannot independently access systems or perform work on the user's behalf. Examples: Microsoft Copilot Chat, ChatGPT, Claude, Gemini.

2. AI Agents (Autonomous AI)

AI agents go beyond conversation by performing tasks and taking actions. These systems can interact with applications, websites, files, or workflows to accomplish a goal with limited human intervention. Typical uses include:

  • Scheduling meetings
  • Managing email workflows
  • Gathering information from multiple sources
  • Running business processes
  • Interacting with software applications
  • Automating repetitive administrative tasks
  • Executing multi-step workflows

Because AI agents may access organizational systems, data, and applications, they introduce additional security, privacy, and compliance considerations. Organizations should carefully review permissions, monitoring, and approval requirements before allowing agent-based automation. Examples: Microsoft Copilot Actions, autonomous workflow agents, task automation platforms, AI-powered robotic process automation (RPA).

3. Custom AI Solutions (Building AI for Others)

Organizations may also create their own AI-powered applications, chatbots, assistants, or services. In this case, AI is not just being used by an individual; it is being developed, configured, or deployed for use by students, employees, or the public. Typical uses include:

  • Student support chatbots
  • Employee knowledge assistants
  • AI-powered search systems
  • Custom copilots connected to organizational data
  • AI-enabled web applications
  • Department-specific AI tools
  • Specialized research or instructional assistants

These systems require the highest level of governance because they may process institutional data, serve large numbers of users, and represent the organization publicly. Additional review is typically required for security, privacy, accessibility, data governance, and ongoing maintenance. Examples: A student services chatbot, a campus-specific Copilot, an AI advisor integrated into a college website.

AI Policies

Skagit Valley College maintains separate guidance for each level of AI use.

  1. This Article summarizes our AI Chat Policy for conversational AI tools used by individuals. The full Policy document is Skagit Valley College Artificial Intelligence Core Policy - Full acceptable use and risk classification
  2. Get permission and guidance for AI Agents and custom AI Solutions from the IT Helpdesk, or contact Brian Culver.
  3. See also: Washington State Artificial Intelligence Policy (WaTech DATA-04)

Quick Decision Guide

If you want to...

 Allowed?

Approval Required? (see IT Helpdesk)

Brainstorm, outline, or get writing feedback

✅ Yes! OK with any AI tool

 No

Get help with code or technical problems

✅ Yes! OK - don't share proprietary code

 No

Summarize a public document or article

✅ Yes! OK with any AI tool

 No

Work with student grades or records

⚠️ Use approved enterprise tools only

 No

Analyze unpublished research data

⚠️ Use local LLM or approved tool only

 No

Draft emails with employee/student names

⚠️ Remove PII before using public AI

 No

Submit AI-generated work for a class

⚠️ Ask instructor; always disclose AI use

 No

Use AI during an exam

❌ Not allowed unless explicitly permitted

 Yes
Give AI access to information available only with a password ❌ Not allowed unless explicitly permitted Yes
Create a chatbot for students to use ⚠️ Ask Director of IT; always disclose AI use Yes

Do and Don't

DO Use AI for:

Do NOT Use AI for:
Brainstorming ideas, summarizing articles, and outlining content Inputting sensitive or private information (student records, health data, SSN, financials, research data)
Generating code scaffolds or pseudocode for programming tasks Submitting AI-generated content as original academic work without disclosure
Administrative or academic support (e.g., organizing notes, rewriting text) Using AI to impersonate others or fabricate data/results
Learning assistance and tutoring (with proper disclosure) Uploading FERPA-protected student data to public AI tools
Drafting emails, templates, and non-sensitive documents Using AI for hiring, admissions, or grading decisions without human oversight
Research ideation and literature review assistance Uploading unpublished research to public AI tools

Back to top

Data and Privacy Caution

Never share Personally identifiable information (PII) with an AI. Do not enter student records, health data, financial information, or confidential research such as:

  • Student names with grades
  • Student ID numbers
  • Social Security numbers
  • Medical information
  • Payroll information
  • Personnel evaluations
  • Passwords
  • Security procedures
  • Incident investigation details
  • Unpublished research data
  • Internal legal communications

Assume anything you enter into public AI tools may be stored, viewed by third parties, and potentially used for model training. Never upload documents containing grades, health info, financial records, or unpublished research to public AI.

Back to top

Academic Use Guidelines

For Students

  1. Assignments: & Essays: Disclose if AI tools were used. Example: "Portions of this response were drafted with assistance from ChatGPT."
  2. Exams & Quizzes: Use of AI tools is NOT permitted unless explicitly allowed by the instructor.
  3. Learning vs. Submitting: Use AI to assist with studying and practice, but do not submit AI-generated work as your own.
  4. When in doubt: Ask your instructor whether AI is allowed and how to cite it.

For Faculty

  1. Transparent Expectations: Clearly communicate AI policies in your syllabus and assignments.
  2. Model Ethical Use: If you use AI to generate rubrics, content, or materials, consider disclosing it.
  3. AI Detection Caution: AI detection tools have significant false positive rates. Do not rely solely on detection—require disclosure instead.
  4. Assignment Design: Consider assignments that incorporate AI appropriately or that AI cannot easily complete.

For Researchers

  1. Protect Research Integrity: Never input unpublished or proprietary research into public AI tools.
  2. Methodology Disclosure: If AI tools are used for analysis or writing, explain the role clearly in your methods.
  3. IRB Considerations: Consult your IRB if AI tools are used in human-subjects research.
  4. Citation: Follow emerging APA, MLA, Chicago guidelines for citing AI assistance.

Back to top

Citing AI Usage

Use footnotes, parenthetical notes, or appendices to describe how and where AI assisted your work.
Examples:

  1. "Drafted outline with assistance from ChatGPT (GPT-4) on March 12, 2025."
  2. "Code debugging assistance provided by Claude. Human review and testing performed before submission."
  3. "Initial literature search conducted with Gemini; all sources independently verified."
  4. APA Style (example): OpenAI. (2025). ChatGPT (Mar 2025 version) [Large language model]. https://chat.openai.com

Back to top

Prompt Engineering Best Practices

  1. Be Specific: Detailed, context-rich prompts improve output quality. Example: "Summarize this policy for a student audience in under 200 words."
  2. Avoid Bias: Don't include prompts that assume discriminatory views or ask AI to generate opinions on sensitive matters.
  3. Iterate: AI outputs often need refinement. Adjust your prompt based on initial results.
  4. Safe Prompting: Never include names, PII, unpublished research, or confidential information in prompts to public AI.
  5. Verify Outputs: AI can hallucinate facts. Always verify important information from authoritative sources.

You are responsible for...

AI can and does make mistakes. AI is not the same as a search engine. AI systems hallucinate, invent citations and references, misinterpret prompts, provide outdated information, and provide biased or incomplete responses. You are responsible for the results. Be aware of you accountability when using AI:

  1. You are accountable for Hallucinations: AI-generated content can be inaccurate, biased, or completely fabricated. Always:
    1. verify facts.
    2. review outputs
    3. follow College policies
    4. Make final decisions
  2. You are accountable for Bias: AI may reflect biases in training data. Always:
    1. Review outputs for fairness, especially in sensitive contexts.
  3. You are accountable for Copyright Issues: AI-generated content may inadvertently reproduce copyrighted material. Always:
    1. Review before submitting to an instructor
    2. Review before publishing.
  4. You are accountable for Accessibility: When creating content with AI assistance, ensure outputs meet accessibility standards (WCAG 2.1). AI can help generate alt-text for images, captions for videos, and accessible document formatting.
  5. You are accountable to Report Concerns: Report harmful, concerning, or policy-violating AI outputs to your instructor or IT/security team.

Back to top

Related Policies & References

Authority

All computer systems at Skagit Valley College are state property. Even if you are using your personal device, by connecting to the Wi-Fi, to Canvas, or to any other system provided by the College, you are using state property. State law requires everyone who uses state property to follow certain laws. Here is the chain of authority governing the College's IT policies:

Back to top

Print Article