Generative AI Policy

To download a copy of this policy, click here.

www.impactfund.org

Protocol for the Use of Generative Artificial Intelligence in Operations and Program Delivery

Effective Date: August 1, 2025

1.       Statement of Purpose

This Protocol is intended to help Impact Fund employees use approved generative artificial intelligence (“AI”) tools with thoughtfulness and care to safely increase operational and program effectiveness.

Generative AI tools allow users to enter prompts to receive human-like text, images, sound, or videos created by artificial intelligence. They can also summarize content, categorize information, and perform numerous other tasks. Examples include ChatGPT, Google’s Bard, and Microsoft Bing.

2.       Factual Background

The field of generative AI is changing rapidly. The following facts are fundamental to this Protocol:  

·       Generative AI tools are prone to “hallucinations,” the creation of false answers or information, and the use of stale information.

·       Generative AI tools rely on patterns of information and can replicate preexisting bias, including implicit and explicit biases that further systems of oppression that the Impact Fund is affirmatively working to eliminate.

·       Information entered into publicly available generative AI tools is shared publicly, which may violate confidentiality, attorney-client privilege, and work product protections, as set forth in the Impact Fund’s Information Security Protocol and Employee Policy Handbook.

·       Early studies indicate that overreliance on generative AI tools impedes learning, critical thinking, creativity, analysis, information retention, skill development, and independent problem solving. Routine use decreases self-confidence and increases reliance on generative AI tools.

3.       Guiding Principles

The Impact Fund adopts the following foundational principles to guide its use of generative AI tools for organizational purposes:

a.       Use of generative AI tools at the Impact Fund will be human centered. Our work is rooted in authentic human ingenuity and connection, therefore our colleagues and constituents must perceive all Impact Fund communications and work product as coming from a person. Misuse of generative AI tools can damage our internal trust in each other and external trust in our work and professional reputation.

b.       Generative AI tools are not a substitute for human judgment and creativity.

c.        Employees hold ultimate responsibility for producing reliable work product via replicable means, even when using generative AI tools.

4.       Acceptable Use of Generative AI Tools

DO:

a.       Use only generative AI tools that have been approved by the Impact Fund for use.

b.       Be familiar with the generative AI tool’s policies on data retention, data sharing, and self-learning.

c.       Undertake training provided by the Impact Fund on proper use of generative AI tools.

d.       Keep your team lead informed on the ways you use generative AI tools in your day-to-day work and obtain advance approval before using generative AI tools for specific projects. Team leads may also specify tasks or projects that are inappropriate for the use of generative AI tools.

e.       Review and verify all generative AI outputs before use, including by identifying and reviewing primary sources and by fact-checking AI outputs against independent research.

f.         Review generative AI outputs for algorithmic (inherited) discrimination, including the perpetuation or recreation of bias.

g.        Maintain the privacy of the Impact Fund’s intellectual property and its Confidential, Restricted, or Privileged/Protected Information (as defined by our Information Security Protocol) at all times.

h.       Comply with all Impact Fund policies, applicable laws, and vendor terms of use.

DO NOT:

a.       Use any generative AI tool for any purpose that could (1) negatively impact the trust that our clients, grantees, and constituents have in us, or (2) endanger the perceived authenticity or reliability of our communications or work product.

b.       Represent work generated by a generative AI tool as your own original work.

c.       Place the Impact Fund’s intellectual property or its Confidential, Restricted, or Privileged/Protected Information (as defined by our Information Security Protocol) into any generative AI tool.

d.       Use generative AI tools for legal or other substantive research without reviewing all primary sources and independently confirming the accuracy of outputs.

e.       Use “self-learning” generative AI tools, unless they have been modified to not store or repurpose user-provided data.

f.         Use generative AI tools to attempt to re-identify anonymized data.

g.        Use generative AI tools to create visual or written content that replaces the ordinary hire and compensation of artists.

h.       Violate the intellectual property rights of authors or visual artists by requesting outputs in their specific style.

i.          Use generative AI tools provided by the Impact Fund for personal gain, unlawful activities, or any purpose that could damage the Impact Fund or our community.

j.          Integrate any generative AI tool with internal software without specific written permission from your team lead and the Deputy Director.

5.       Generative AI Tools Approved for Use

Employees may only use generative AI tools previously approved by the organization. To be approved, technologies must protect data from public disclosure and unauthorized access, including through anonymization, encryption, and secure storage practices.

A list of approved technologies is available upon request. Employees are invited to recommend new tools to the Deputy Director for approval.

6.       Violations

Maintaining this Protocol is considered part of each employee’s job duties. Violating this Protocol may result in disciplinary action as described in Section 2.15 (“Performance Improvement & Development”) of the Employee Policy Handbook. Disciplinary action may include suspension, access restrictions, work assignment limitations, or more severe penalties up to and including immediate termination, in accordance with applicable law. Suspected violations of this Protocol should be reported to the HR Director or the Executive Director.

7.       Disclosure

The Impact Fund is transparent in its use of generative AI tools internally and externally to encourage trust in our work. A copy of this Protocol is publicly available on our website, www.impactfund.org.

# # #