SHI’s AI Policy

Created July 2026

Note: We created this policy after a presentation by Jane Franch on the Responsible Use of AI. It is a work in progress that we will continue to update and modify as we learn more.

SHI Mission

SHI’s mission is to create a movement supporting high quality herbal products, sustainable and ethical sourcing, and greater overall transparency in how herbs reach end users.

Generative AI offers a chance to help nonprofit organizations like Sustainable Herbs Initiative in new ways. AI has the capacity to assist and/or expediate certain tasks that are part of our workload such as editing or conducting research review. This guidance document is an attempt to outline some pitfalls and best practices when using AI at SHI.

As an overarching principle, we bring the same questions and values to our use of generative AI that we bring to all of the work at SHI: Where does it come from, who built it (and whose voices aren’t included), what is the full cost (on the energy, climate, water, people), and who bears those costs now and in the future?

AI is shaped by who uses it and what for. We have written this policy to ensure that we use these tools with intention and awareness of their impacts. And we commit to using our voice to demand clean infrastructure, data transparency, and structural change.

Scope: This policy applies to all SHI staff and any contracted work. We also share this as a template for SHI member companies to consider in developing their own AI policy.

Guidelines

1. Humans in the loop, all outputs reviewed before using them.

• AI will never be used to simulate or substitute for direct human relationship with SHI members, producers, wild harvesting communities, or the plants themselves.
• Any AI content, edits or idea should always be thoroughly vetted (with sources in the case of the research) and reviewed (in the case of any writing) by the employee before any final product is presented. If appropriate, staff should ensure people know what parts of their work were generated by AI.

2. Usage boundaries

• Favor standard search engines for simple queries.
• Reserve AI for situations where it brings real added value, freeing up time for the deeper thinking, relationship building, and facilitation work that is SHI’s core function. Potential uses include: thought partnership on complex problems, research and summarizing research, help drafting and/or editing administrative work, summarizing key points from meetings, editing more in depth writing.
• Use text generation prompts and occasional graphic help.
• Restrict very high energy-intensive uses (images and video generation).

3. Cognitive offloading and the protection of thinking.

• Use AI to free up human judgment, not replace it. Before using AI for any task, ask: am I reaching for this because it genuinely helps or because I’m avoiding the discomfort of thinking/working through it myself?

4. Build AI literacy.

• Since AI is rapidly evolving, we will continue to educate ourselves and SHI members about the impacts of AI use and share resources that help minimize the footprint of our use and maximize the positive handprint (impacts) of its use overall.

5. Address climate impact and environmental questions

• Identify tools that can help companies measure climate impact and reduce their emissions.
• Explore measuring carbon footprint of AI use via EcoLogits Calculator
• Research and experiment with lightweight open-source models (e.g., Mistral 7B)
• Avoid repeated unnecessary prompting;
• Avoid image/video generation unless clearly needed;
• Reuse summaries and prompts where appropriate;
• Prefer tools/vendors with renewable-energy, water-use, or emissions transparency;
• Continue to check Sustainable AI Group for updates and guidance on how SHI can help push the AI industry toward a version that works for people, planet, and profit.

6. Training data, knowledge bias, and whose knowledge counts

• AI training data skews toward dominant languages, cultures, and commercial interests. Indigenous and traditional ecological knowledge is underrepresented or misrepresented.
• Commit to understanding the digital access of herb producers and processors and identify ways that SHI can address those divides.

7. Privacy and data protection

• Treat AI prompts as you would a public document — don’t put in anything you wouldn’t want shared.
• More specifically do not enter the following:
  • Personal data about members, donors, employees, contractors, producers, or community partners;
  • Financial, legal, HR, health, or personnel information;
  • Passwords, credentials, contracts, grant details, and internal strategy.

     8. Governance and Transparency

• Clearly inform when AI is used in content or services
• Explain tool choices and their implications
• Share learnings with other sector actors

9. Cost

• Evaluate AI tools not just for their direct cost but for their full cost: what human capacity they free up, what they might quietly erode, and whether the environmental cost is proportionate to the benefit?

10. Intellectual property and attribution

• AI outputs draw on existing writing and knowledge without attribution. We are committed to clear attribution of sources.

11. Accountability

• Concerns about the use/misuse of AI will be brought to the SHI Stewardship Council to discuss.

Revisit this policy annually, share it publicly on the SHI website, and be transparent about how AI is and isn’t used in SHI’s work.

Questions or reflections on this policy are welcome at ann@herbsinitiative.org.

Sources

This policy was developed based on the guidance in Beyond the Hype: A Sustainability Professional’s Guide to Shaping the AI Future Presentation to the SHI June 2026 Monthly Meeting, Jane Franch, AI for Earth Ambassador, Climate Collective, June 2026. It is modeled on the A Charter Designed for the BeEducation Network and on the NTEN Responsible AI Use Policy framework with input from Claude.