AI & Social Media Policy
How Diaspora Peoples Place Uses AI and Social Media
Diaspora Peoples Place is a volunteer-run 501(c)(3). We have no paid staff, and in 2026 we leaned on AI tools to carry the behind-the-scenes work of running an organization. Many in our community have serious, well-founded concerns about AI, and we share many of them. We feel the same way about social media platforms that profit from our community's attention, data and creative work. We are publishing this page so you can see exactly how we use AI and social media, what we estimate AI costs the planet, and the limits we hold ourselves to.
The concerns we take seriously
Frontline communities and the environment
AI runs in data centers that draw heavily on electricity and water. Those facilities are often built in communities with the least power to refuse them, which then absorb the pollution, the strain on water supplies and the rising utility bills. Even in California, data centers are disproportionately sited in marginalized neighborhoods already burdened by industrial pollution, and here in the San Gabriel Valley, construction on a proposed Monterey Park data center was halted in February 2026 after community outrage and a lawsuit. We see this as an environmental justice issue, not only a carbon issue.
Artists and creative labor
Image generators and similar tools were built largely on artists' work without consent or pay, and working creatives are feeling it. In the Association of Illustrators' 2026 survey, 46% of respondents knew of a client who chose generative AI over them, up from 32% a year earlier. DPP exists to uplift diaspora creatives, so we will not use AI in ways that take work or credit from them.
Jobs
When organizations use AI to avoid hiring people, workers lose. Our AI use covers work that a volunteer-run organization could not otherwise pay anyone to do, and it never replaces paid creative work.
Social media and consent
We make graphics for social media, but we know these platforms extract value from creatives. Meta trains its AI on public Facebook and Instagram posts, and users in the United States have never been given a way to opt out. In July 2026, Meta's Muse Image tool briefly allowed its AI to reference public Instagram accounts to generate images until public backlash forced a reversal. Anything posted publicly on Instagram, including an artist's work that we share, can feed those systems. This is one reason we don't emphasize original creative work on social media, and why we are moving our events calendar and community updates to our own website and email list.
How we used AI in 2026
Our AI use fell into three areas.
- Website and digital infrastructure. AI helped us write the code for our Greater LA events guide on diasporapeoplesplace.org, gather community event flyers for it, and fix technical issues on our site.
- Administration. AI helped us build budgets, financial trackers and reimbursement logs, draft partnership agreements, and organize our nonprofit compliance calendar with our accountant.
- Social media visuals. AI helped us create text-based social media posts, weekly events calendar carousels and cover graphics. These are assembled from real event flyers, real photographs and our own design elements. We do not use image generators to create artwork.
AI is not the author of our storytelling, our journalism or the art at our events. Performances, poetry, music, photography and our voice remain human work.
Our estimated AI footprint
AI companies, including the one whose tools we used most, do not publish energy or water data for their models, and we do not have usage logs that would allow a precise audit. The estimate below is our best reconstruction. We reviewed our 2026 work records to count substantial AI work sessions by area, then applied published research on the energy use of multi-step AI sessions, which ranges from about 40 to 600 watt-hours per session. Because much of our work involved long sessions on a large model, we assumed roughly 100 to 430 watt-hours per session. We converted energy to water and carbon using commonly cited averages of 2 milliliters of water and 0.42 grams of CO2 per watt-hour.
Our 2026 AI use was about one day of one home's electricity
An average US home uses about 10,950 kWh of electricity a year, roughly 500 times our whole year of AI use.
Our central estimate is about 22 kWh of electricity, 45 liters of water and 9 kilograms of CO2 for the year. For scale, an average US household uses roughly 30 kWh of electricity per day, so our year of AI use is comparable to about one day of electricity for one home. That is small next to driving to a single event, but it is not zero, and the communities near data centers are the ones who pay it. These figures carry real uncertainty and may be off by a factor of two in either direction.
Our AI policy
Several of these commitments are adapted from published AI policies and guidance from other organizations, including Candid, Whole Whale, The Green Web Foundation, the NAACP and Bridgespan.
1Keep our footprint low
- Ask first whether AI is needed. If a volunteer, a template or a simpler tool can do the job, we use that instead.
- Use the lightest model that works. Routine tasks go to small, efficient models, which can use a small fraction of the energy of the largest ones. Large models are reserved for complex builds.
- Cap heavy sessions. We limit long, multi-step AI sessions to four per month and group related tasks into one session instead of many.
- Set an annual target. Our goal for 2027 is to keep our AI use under 15 kWh, roughly a third below our 2026 estimate, and we will report how we did.
- No AI image, video or voice generation. These are among the most energy-intensive uses, and they draw directly on creative labor.
2Protect artists and creative work
- We never use AI to create or imitate the work of artists, performers, poets, photographers or vendors.
- We never present AI-generated images as real photographs, and we do not create synthetic voices or likenesses of real people.
- AI never replaces a paid creative commission. When our budget allows design, photography or video work, we hire people.
- We share artists' original work on social media only with their permission, credited and linked, and we tell them that public posts on Meta platforms may be used for AI training.
- Before we use a culturally specific motif or symbol in our branding or merchandise, we consult members of the community it comes from.
A note on our community's flyers. Many community organizations and diaspora businesses we feature use AI to make their flyers, often for the same reason we use it for our back office: no budget and no staff. These commitments govern our own work, not theirs. We do not exclude events from our calendar or judge partners based on how a flyer was made. Where we can, we connect organizers with diaspora designers and illustrators in our network, because the best way to protect creative work is to keep it paid.
3Protect our community's data
We do not put donor financial information, personal details about vendors, volunteers or children, or private legal matters into AI tools. We use account settings that keep our inputs out of model training wherever those settings exist, and we maintain a short list of approved tools that we review twice a year.
4Keep people accountable
A person reviews and verifies everything AI helps produce before it is published or sent, and AI never makes decisions about vendor selection, artist fees or how funds are spent. No volunteer is ever expected to use AI, and anyone who has questions about a use can pause and ask without penalty.
5Be transparent
This page is public, and we update it each year with a new footprint estimate. When a funder asks, we disclose whether AI assisted with a grant application.
6Give back and push for accountability
Each year we make a contribution to a local water or environmental justice organization in the San Gabriel Valley or Greater LA, and we name it in our annual report. We chose a local contribution over buying carbon offsets because many offset markets are hard to verify, and a local contribution is one our community can see. We also stand with frontline communities organizing for transparency and accountability from data center developers, including efforts guided by the NAACP's Frontline Framework.
Our social media policy
Social media is where much of our community finds out what is happening, so we have not left it entirely. But we treat it as a doorway, not a home. The platforms profit from our community's attention, data and creative work, and they can change the rules at any time. Our aim is to use them only as much as it takes to stay connected to local events and to bring people into spaces our community controls.
1Use social media as a doorway
Our posts point people toward our website events calendar, our email list and our events, where we can reach our community without an algorithm or an ad platform in between. Supporters who give $5 a month receive our community events calendar in their inbox twice a month, which helps fund our move off the platforms.
2Post less, and post with purpose
We post only event flyers, recaps of our events and our weekly community events calendar. After a few weeks we archive older flyers and calendars, so our profile shows our impact, our next event and the latest calendar without building a permanent library of content for the platform. Three pinned posts tell people who we are and where to find us off the app.
3Protect creative work
We do not emphasize original creative work on social media. We feature artists, performers and vendors by promoting their events and linking to their own websites and shops rather than posting their portfolios. We share an artist's original work only with their permission, and we encourage creatives to keep the full version of their work on platforms they control.
4Protect the people in our photos
We ask for consent before posting identifiable photos of community members, and we honor every request to be left out or taken down. We do not post identifiable photos of children without a parent's or guardian's permission. At protests, vigils and other politically sensitive gatherings, we avoid showing faces that could put people at risk.
5Stay informed and give credit
We follow community organizations, venues and diaspora businesses so we can keep up with local events for our calendar. When we share an event, we use the flyer its organizers already made public, credit the host and link to their account, and we correct or remove a listing whenever an organizer asks.
6Limit what the platforms can take
Wherever a platform offers a setting that keeps our posts out of AI reuse, we turn it on, and we submit Meta's AI training objection request even though it is not guaranteed for US users. We do not collect sign-ups, donations or personal information through direct messages. We send people to our website, where their information stays with us, and we never sell or share our email list.
Review
Our founder and board review these policies every November. If you have a concern or an idea about how we use AI, we want to hear it.
Sources
- Association of Illustrators 2026 survey, via Resultsense
- Jiang, Taylor and Agnew, How Professional Visual Artists Are Negotiating Generative AI in the Workplace (2026)
- Computerworld: Meta and US users' data for AI training
- Crypto Briefing: Meta reverses course on Muse Image
- Center for Health Journalism: The AI data center boom and community health
- University of Chicago Sustainability Dialogue: Data centers, pollution and the communities left behind
- NAACP: Holding data centers accountable
- Zeke Hausfather, The real energy use of agentic AI (via Skeptical Science)
- Simon P. Couch, Electricity use of AI coding agents
- Agentic AI energy and water estimates (Couch-based analysis)
- Projet Celsius: Claude's carbon footprint by model
- Candid: Responsible AI use policy for nonprofits
- Whole Whale: The Nonprofit AI Toolkit
- Bristol Climate & Nature Partnership: The Green Web Foundation on AI emissions
- Nonprofit Quarterly: Choosing Your AI Path