A Note Before You Start

This template is a starting point, not a legal document. Have your legal counsel review the final version before adopting it — particularly the sections on data privacy and beneficiary information.

This policy matters most for organizations that serve vulnerable populations: children, undocumented individuals, domestic violence survivors, people experiencing homelessness, people with mental health or substance use histories, or anyone whose data, if disclosed, could cause them harm.

For organizations whose work does not involve sensitive beneficiary data, a shorter, lighter version of this policy is appropriate. The template includes guidance on which sections to prioritize based on your situation.

[ORGANIZATION NAME] ARTIFICIAL INTELLIGENCE USE POLICY

Adopted: [Date] | Version: 1.0 | Next Review: [Date, recommend annual]

Section 1: Purpose

This policy establishes how [Organization Name] staff, volunteers, and contractors may use artificial intelligence (AI) tools in their work. It is designed to protect the privacy and dignity of the people we serve, maintain the integrity of our donor and funder relationships, and ensure that AI use aligns with our organizational values.

AI tools can meaningfully increase our team's capacity. They also carry risks that require deliberate management. This policy does not prohibit AI use — it provides a framework for using it responsibly.

Section 2: Scope

This policy applies to all staff, volunteers, and contractors who use AI tools as part of their work for [Organization Name], including but not limited to:

• AI writing and editing tools (Claude, ChatGPT, Gemini, and similar)
• AI data analysis and summarization tools
• AI tools embedded in platforms we use (CRM AI features, email AI assistants, etc.)
• Any future AI tools adopted by the organization

Section 3: Approved Tools

The following AI tools are currently approved for use at [Organization Name]:

Staff who wish to use an AI tool not on this list must request approval from [role/name] before use. Approval requires a brief written summary of the intended use case and a review of the tool's data terms.

Section 4: What You May Use AI For

Staff may use approved AI tools for the following:

• Drafting and editing written communications, grant applications, reports, and internal documents
• Summarizing meeting notes, research documents, and other organizational materials
• Generating ideas, outlines, or structures for content that a human will review and finalize
• Analyzing organizational data that does not include personally identifiable information (PII)
• Automating repetitive tasks that do not involve sensitive data

Section 5: What You May Not Use AI For

Prohibited uses:

• 5.1 — Do not input individually identifiable beneficiary information into general AI tools. This includes names, contact information, case numbers, diagnoses, immigration status, legal history, or any other information that could identify a specific person we serve.
• 5.2 — Do not use AI-generated content in any external communication without human review. AI output must be edited and approved by a staff member before it is sent to donors, funders, the media, or the public.
• 5.3 — Do not use AI to fabricate, estimate, or extrapolate program data or statistics. Any numerical claim in a grant application, impact report, or donor communication must be verifiable against our actual records.
• 5.4 — Do not use AI to replace direct human contact in situations that require it. This includes: responding to beneficiaries in distress, managing sensitive donor relationships, and any situation where a community member expects to be communicating with a person.
• 5.5 — Do not use AI tools to make decisions about program eligibility, hiring, or resource allocation without human review and final authority.

Section 6: Handling Beneficiary Information

This section applies to all organizations serving individuals with sensitive data. If your organization does not collect or use individual-level beneficiary data, this section may be simplified.

6.1 — Anonymization before AI use

If you need to use client or beneficiary information in an AI tool (for example, to analyze survey responses or draft impact stories), you must first anonymize the data. This means removing: names, addresses, specific dates, case identifiers, and any detail that could identify an individual.

Replace individual identifiers with general descriptors: "a 34-year-old single parent" rather than a name; "a client served in our East Side program" rather than an address.

Section 7: Quality Control and Human Oversight

7.1 — The review requirement

All AI-generated content used externally — grants, donor communications, reports, social media, press materials — must be reviewed and approved by a designated staff member before release. The person who requested the AI output may not serve as their own sole reviewer for high-stakes materials (grant applications, major donor communications, public statements).

Section 8: Data Security

8.1 — Account security

AI tool accounts used for organizational work should use strong, unique passwords and, where available, multi-factor authentication. Do not use personal AI accounts for organizational work.

Section 9: Disclosure

9.1 — Grants and funder communications

Some funders are beginning to ask about AI use in grant applications. When asked, be honest. Disclose that AI tools are used for drafting and that all final content is reviewed and verified by staff. Do not claim that work is entirely human-produced if AI was used.

Section 10: Policy Review and Updates

This policy will be reviewed annually and updated as needed when:

• The organization adopts a new AI tool
• Significant changes occur in relevant laws or sector standards
• An incident occurs that reveals gaps in this policy

Staff with questions about this policy should contact [role/name].