Echidna's five organizational principles, ratified with the Board in 2022 out of the Sacred Cow discussions. They're reproduced here because the AI policy is being built directly on top of them: every AI rule we write should trace back to a line on this page. The full extended document, with how each principle shows up in grantmaking and internally, lives in the shared Drive doc Emily circulated.
Core function is making grants: multi-year by default, least restricted form that aligns with strategy, with support beyond grant dollars (networks, advice as requested, best practices, field resources). Comfortable that supportive grantmaking sometimes looks "boring" by traditional metrics. Approaches to philanthropy are shared as public goods.
The AI policy itself becomes one of those public goods: shared with grantees and other funders once it exists.
Local leadership first, grantees take the lead, contribution without control. Evidence taken broadly: empirical evidence and lived experience both count. Internally: all employees comfortable expressing opinions and influencing decisions; impact prioritized above fixed procedures.
Staff shape the tool decision through the diagnostic, and AI never substitutes for hearing directly from grantees and communities.
Time invested in understanding grantees and their context. Respect for people's time: do the homework, ask only for what gets used, avoid bureaucracy, take pride in responsiveness. Materials that grantees share get read and reacted to, in acknowledgment of the effort involved in keeping Echidna informed. Worries get raised with grantees directly, and no comes as soon as it's known.
The anchor principle for the policy, named in the room on July 29. Reading grantee materials personally is how Echidna honors grantee effort, so the read-first rule sits here. AI that speeds responsiveness serves this principle; AI that substitutes for engagement breaks it.
Feedback sought continually, as an organization and as individuals. Grantees taking measured risks is valued even when they sometimes fail; the response to failure is measured, because what matters is learning from it. Internally: no innovation for its own sake, and funds exist for training and career development.
The permissive-with-guardrails posture is this principle in action: a learning experiment with a checkpoint, blameless incident reporting, and no tool adopted just because it exists.
Diversity in hiring, decision-making, and culture is a priority. Funding and power correlate, so Echidna questions its own power and privilege and works to cede it to the communities it hopes will thrive, correcting for its biases as a US-based organization.
Training covers AI bias directly, with attention to how models represent the communities Echidna serves, and the same self-examination applies to what AI outputs get trusted.
Kim asked in the diagnostic that we review the principles as part of building the AI guidance, and Emily shared the extended document after our July 29 session. The policy pre-read does the detailed mapping from these principles to draft AI rules; this page is the reference it maps from. If the policy ever asks the team to do something these principles wouldn't, the policy is what needs to change.
Principle text condensed from the Echidna Giving Principles (final, 2022). The extended version, including how each principle shows up in grantmaking and internal operations, is in the Drive document shared by Emily on July 29.