CAOS AI Policy

Purpose

CAOS is not an AI organization, but AI now touches everything and everyone — including our mission. As it applies to us, we support open-source AI development that is created and trained in close collaboration with people with disabilities, treats accessibility and usability by disabled people as a non-negotiable requirement, and actively works to avoid creating or reinforcing ableism and other disability-related inequities. We also encourage the ethical use of AI in assistive technology specifically — welcoming its real accessibility benefits while insisting on reliability, quality, user control, privacy, and appropriateness.

These are organizational positions, not commitments to specific advocacy or programmatic work. How much, and in what form, CAOS engages publicly on AI and accessibility is a question for our strategic planning process, not this policy.

This policy covers the other half of that picture: how CAOS staff, board members, and contributors use AI in their own work on behalf of the organization.

Scope

This policy applies to all board members, staff, volunteers, and contractors acting on behalf of CAOS, and to any AI tool used in CAOS governance, operations, communications, or programs.

Policy Statement

  • AI is a tool, not a default. We use it with intension, and only when it offers significant efficiency or quality benefits.
  • A human is always in the loop. We never publish or act on unvetted AI-generated content.
  • We choose tools aligned with our mission. We favor AI tools and models with credible accessibility training, meaningful disability awareness, and open-source values.
  • We recognize environmental impacts. We select tools appropriate in scale to the task, minimize our AI usage, and avoid using models known to be at the deep end of environmental cost.

Responsibilities

Board of Directors

  • Adopt and maintain this policy
  • Approve the AI Tooling Memorandum and any material changes to it

Executive Director / Board President

  • Maintain and publish the current AI Tooling Memorandum
  • Ensure contributors know this policy and the Memorandum exist and apply to their work

Staff, Board Members, and Contributors

  • Follow this policy and the current AI Tooling Memorandum in all CAOS-related work
  • Propose changes to the Memorandum through the usual issue and discussion process

Procedures

AI Tooling Memorandum

The Board maintains a standing AI Tooling Memorandum — practical guidance on which AI tools are approved for CAOS work, and how they should and shouldn’t be used. Tool-specific detail lives in the Memorandum, not in this policy, so this document doesn’t need revising every time the AI landscape shifts.

Anyone can propose an addition, removal, or change to the Memorandum by filing an issue in the governance repository. The Board reviews and updates the Memorandum as needed.

Revision History

Version Date Summary
0.1 2026-06-29 Initial bulleted draft
0.2 2026-06-30 Expanded prose draft
0.3 2026-07-01 Rewritten for conciseness ; tool-specific detail moved to AI Tooling Memorandum mechanism