AI Resources

As agency leadership, you are accountable for all AI implementations within your organization. This means making final approval decisions, ensuring adequate resources, and maintaining strategic oversight. You work in partnership with ITS, but the responsibility for appropriate AI use within your agency rests with you.

Idaho’s shared responsibility model positions you as the decision authority, while ITS provides the framework, standards, and consultation. Your IT liaison handles day-to-day operations, but you own the outcomes.

Every AI system requires your approval based on Idaho’s three-tier risk classification:

• Tier 1 (Low Risk) – Systems using public information with minimal impact require your IT leader’s approval in addition to ITS approval. These included basic chatbots or content assistance tools using only public data.

• Tier 2 (Medium Risk) – Systems with moderate impact or sensitive data require your direct approval plus ITS review. Examples include document analysis systems or recommendation engines that influence operations.

• Tier 3 (High Risk) – Systems affecting individual rights or using critical data require AI Executive Committee approval after your endorsement. These include eligibility support systems or anything impacting citizen services.

The key question for your approval: What happens if this system fails or makes a mistake? The answer determines the risk tier and approval requirements.

• Start strategically – Choose low-risk pilots that deliver clear value and build organizational confidence. Success with simple applications creates momentum for more complex implementations later.

• Budget appropriately – AI governance requires investment in training, documentation, security controls, and ongoing monitoring. Higher-risk systems need more resources. Factor these costs into your planning.

• Engage ITS early – Involve ITS during concept development, not after decisions are made. Early consultation prevents costly redesigns and ensures alignment with state standards.

• Think beyond technology – AI impacts workflows, staff roles, and citizen interactions. Plan for change management, training, and communication alongside technical implementation.

Most agencies will procure AI capabilities rather than build them. You are responsible for ensuring vendors meet Idaho’s requirements through proper due diligence and contractual protection. 

Require vendors to complete Idaho’s standardized assessment questionnaire. Include compliance clauses, security requirements, and audit rights in contracts. Establish ongoing oversight procedures for vendor performance and system updates. 

Remember: You remain accountable for AI outcomes even when using vendor solutions.

• Designate clear accountability. Your IT Liaison manages AI operations, but you own strategic decisions and outcomes. Maintain regular oversight without micromanaging day-to-day activities. 

• Monitor what matters. Focus on business outcomes, risk indicators, and citizen impact rather than technical metrics. Review performance quarterly for low-risk systems, monthly for medium-risk systems, and weekly for high-risk systems. 

• Build organizational capability. Invest in AI literacy across your leadership team. Participate in Idaho’s AI community of practice to learn from other agencies. Create feedback loops for continuous improvement. 

• Plan for incidents. AI systems will have problems. Ensure your team knows how to respond quickly and appropriately. Have clear escalation procedures to ITS and other state resources. 

• AI Concept Brief – Use this standardized template to evaluate and document proposed AI implementations before detailed planning begins. 

• ITS AI Innovation Team – Your primary resource for policy interpretation, technical guidance, and coordination with state-level governance bodies. 

• AI Framework and Supporting Documentation – Idaho AI Framework provides overarching guidance, and the AI Policy, Standard, and Guideline provide complete requirements and procedures. 

Success depends on your leadership commitment to responsible AI governance. This means making informed decisions, providing adequate resources, and maintaining accountability for outcomes within your organization. 

If you have additional questions or are ready to get started, contact your ITS Service Delivery Manager.

As a state employee, you can use approved AI tools, including Claude, ChatGPT, and Copilot Chat, to help with your work safely and responsibly. You remain responsible for all work done in your official capacity, even when AI assists you. Think of AI as a helpful tool that requires your oversight, not a replacement for your judgment. 

Before using Claude, ChatGPT, or CoPilot Chat, verify three things:

1. Information type: Use only public information (published reports, websites, public documents) or internal non-sensitive information without personal details (drafts, memos, procedures).
   Never use personal or confidential information (Examples: names with identifying details, Social Security numbers, medical records, and financial data). 
2. Tool approval: Confirm the AI tool is on your agency’s approved list. Contact your IT liaison if unsure.
3. Review commitment: You must review all outputs and mark content as AI-assisted.

The newspaper test: If you wouldn’t be comfortable seeing information on the front page of a newspaper, don’t put it in AI tools.   

For Claude, ChatGPT, and Copilot Chat: 

Account Setup:  

• Use your email for AI tool accounts or create a login  
• Enable two-factor authentication and use strong passwords per state IT requirements  
• Turn off conversation history retention and data use for training (where available)  
• Don’t share account credentials with colleagues  

Session Management:  

• Start each work session with a new conversation  
• Don’t reference previous conversations or ask AI to “remember” past interactions  
• Regularly review and delete conversation history  
• Log out completely when finished, especially when switching between work and personal use  

Data Protection:  

• Understand that individual plans may retain data longer than enterprise versions  
• Never save conversations containing work information  
• Report data retention concerns to your Agency IT Liaison  
• Follow agency-specific data protection requirements

Every time you use AI:  

• Use only approved tools with appropriate information  
• Review and verify all outputs against reliable sources  
• Mark content as “AI-assisted,” “AI-generated,” or similar notation  
• Follow your agency’s templates and style guides  
• Report inappropriate, biased, or incorrect outputs immediately  

Good uses: Draft announcements with review, summarize public documents, convert technical information to plain language, format existing content, research public information (verify sources). 

Never use AI for: Final decisions about citizen services, official documents without review, personal/confidential information, or publishing without verification. 

Contact your Agency IT Liaison immediately for:  

• Tool approval questions  
• Accidentally including sensitive information in AI  
• Biased, incorrect, or inappropriate AI outputs  
• Technical problems or security warnings 

Essential Contacts:  

1. Your Agency IT Liaison: All AI questions  
2. ITS Support: Technical issues and policy guidance  

Key reminders:  

• Only use Claude, ChatGPT, and CoPilot Chat with public or non-sensitive internal information  
• Always review, verify, and mark AI assistance  
• Protect sensitive information by never including it in AI tools  
• When uncertain, ask your Agency IT Liaison 

Resources:  

• Your agency’s approved AI tool list  
• Agency IT Liaison contact information  
• Idaho AI Framework
• AI Governance Policy, Standard, and Guideline

Use your Idaho AI Quick Reference Guide to verify: 

1. The AI tool is on your agency’s approved list 
2. You’re using only public or non-sensitive information 
3. You’re logged in with your email account 

Be specific and show examples. Instead of giving vague AI instructions, be as clear as you would with a new employee.  

• Vague: “Write an email about the policy.”  
• Specific: “Write a 2-paragraph email to county clerks about the new records retention policy. Professional tone. Include the March 15 deadline and link to the full policy on our website.” 

Example 1: Professional Emails 

• Write a [short/medium] email to [recipient] about [topic]. 
• Tone: Professional and helpful 
• Include: [specific details they need] 
• End with: [clear next step] 

Example 2: Document Summaries 

• Summarize this public document in three bullet points for [audience] 
• Main change or announcement 
• Who it affects 
• Required action and deadline 

Example 3: Public Communications 

• Write a public notice about [topic] for Idaho residents 
• Use simple, clear language 
• Include: what’s happening, when, how it affects them, and where to get more info 
• Keep under 150 words 

• You are helping an Idaho state employee write [type of content]. 
• Audience: [who will read this] 
• Purpose: [what you’re trying to accomplish] 
• Tone: Professional and clear 
• Include: [specific requirements] 
• Format: [how you want it structured] 

• Too generic? → Add your specific agency context  
• Wrong tone? → Specify “for internal staff” vs “for public”  
• Missing key info? → End with “Must include [specific details]” 

After AI gives you results:  

1. Review content for accuracy and appropriateness  
2. Verify any facts, dates, or policy references  
3. Mark the content as “AI-assisted” or “AI-generated” per state requirements  
4. Check that it follows your agency’s style and templates  

1. Pick one basic task (like drafting an email)  
2. Use a prompt template from the previous section  
3. Try it with non-sensitive information first  
4. Review and edit the results  
5. Mark as AI-assisted before using  

Contact your Agency IT Liaison if: 

• You’re unsure if information is appropriate for AI 
• AI produces biased or inappropriate content 
• You accidentally included sensitive information 
• You need clarification on approved uses 

Remember: AI helps you write faster, but you’re still responsible for the final content. Your expertise and judgment are essential.  

For advanced techniques and detailed examples, see additional resources below.