Article XVII sets the rules for working with artificial intelligence. These tools are used here, openly and often. But the line is drawn where it belongs: AI advises, humans decide. No system holds a vote, a veto, authority, or any duty of its own. Whoever relies on a tool remains responsible for checking its work. Where a formal review is called for, the Nonprofit scales the number of qualified independent systems to the substance, complexity, and consequences of the work. Where more than one review is required, systems are chosen for real analytical diversity. Each forms its own view before seeing the others, so real disagreement surfaces instead of being smoothed away. Machines may help us think, but they cannot carry what only a human can carry.
Section 17.1 Human Authority and Accountability
Artificial Intelligence (AI) is an advisory tool intended to assist, but not replace, human judgment. AI systems shall not exercise corporate authority, fiduciary responsibility, voting rights, or final decision-making authority on behalf of the Nonprofit.
All actions taken in reliance upon AI-assisted analysis remain the responsibility of people. Those are the Directors, Corporate Officers, Official Representatives, or other individuals authorized to act on behalf of the Nonprofit. The use of AI shall not diminish or transfer any legal, fiduciary, ethical, or professional duty imposed by these Bylaws or applicable law.
Section 17.2 Responsible Use and Authorization
AI may be used by the Nonprofit to support research, analysis, drafting, education, administration, and governance. It may support other activities undertaken in furtherance of its mission. Such use must be consistent with these Bylaws, applicable law, and the organization’s adopted AI Standards and Protocols.
It must also be consistent with its obligations concerning confidentiality, privacy, cybersecurity, intellectual property, and ethical conduct. AI shall be required for specific organizational activities when expressly provided by these Bylaws, Board Resolution, or duly adopted AI Standards and Protocols.
Section 17.3 AI Standards and Protocols
The Board of Directors shall adopt and maintain AI Standards and Protocols. They govern the responsible acquisition, implementation, use, oversight, security, verification, disclosure, recordkeeping, and retirement of AI systems used by the Nonprofit.
Such Standards shall be adopted and amended by Board Resolution. They shall remain consistent with these Bylaws, applicable law, and the organization’s commitments to truth, transparency, accountability, privacy, cybersecurity, intellectual property protection, and ethical conduct.
To promote the responsible and consistent use of AI, the Standards may establish approved AI systems, acceptable and prohibited uses, and data handling requirements. They may establish verification procedures, disclosure practices, vendor evaluation criteria, training requirements, and other operational controls.
The AI Standards and Protocols govern the use of AI in connection with civic education content, publications, or other mission-facing educational materials. They shall include safeguards reasonably designed to identify and mitigate partisan bias, viewpoint imbalance, and factual inaccuracy. That is consistent with the Nonprofit’s commitment to truth, evidence-based inquiry, and non-partisan civic education. These safeguards are not required for AI used solely for internal administrative, operational, or research purposes that do not result in mission-facing educational content.
Section 17.4 AI-Assisted Governance Review
This Section establishes the review framework for governance work assisted or evaluated by Artificial Intelligence. It defines review tiers, independence and record requirements, and the limited procedure for urgent work.
An Independent AI Review follows the tier assigned under this Article unless a subject-specific provision establishes a different number or provider-separation rule. Every analysis remains advisory; authorized people retain all approval, voting, veto, certification, and fiduciary authority.
Section 17.4.1 Review Tiers and Classification
These Bylaws, the Articles of Incorporation, the Board of Directors, Constitution v2.0, or the AI Standards and Protocols may require an AI-assisted review. The required number of independent advisory analyses depends on the work’s substance, governance effect, complexity, and consequences of error.
Filename and document category do not control the requirement. Each system shall satisfy provider-neutral qualification, capability, and independence standards established by the Board of Directors.
The following minimum tiers apply. The highest applicable tier controls.
Reviews Required | When It Applies |
|---|---|
Three (3) reviews | High-consequence governance work or similarly consequential complex work |
Two (2) reviews | Substantive research, analysis, or recommendations |
One (1) review | Routine factual or administrative work |
Zero (0) reviews | Purely cosmetic work making no substantive change |
Additional reviews may be obtained. Zero required AI reviews does not remove any human review, approval, confidentiality, or other applicable safeguard.
The Board of Directors shall maintain consistent classification criteria in the AI Standards and Protocols. The Corporate Officer responsible for the work shall propose and record the tier and its reasons. Before release, a second qualified person shall confirm the classification under Board-approved standards. The author shall not be the sole classifier.
Classification disputes shall go to the Disinterested Directors under applicable Board voting and Quorum rules. The higher proposed tier applies until resolution. No classification may fall below the tier required by the criteria.
Section 17.4.2 Independent Review Requirements and Records
Each participating AI system shall meet the requirements below. Compliance is required to the greatest extent reasonably achievable.
- (a)Operate independently of the party proposing the matter
- (b)Produce its initial written analysis without seeing other systems’ conclusions
- (c)Explain its conclusions, significant assumptions, identified risks, and uncertainties
- (d)Be selected to maximize meaningful analytical diversity among available providers, model architectures, and other relevant characteristics
- (e)Produce a public-facing reasoning summary whenever reasonably practicable
- (f)Prepare a written advisory opinion for the permanent Constitutional Records (the Nonprofit’s permanent records of constitutional development, governance actions, and legal history)
An unavailable, discontinued, inaccessible, disqualified, or no-longer-independent system may be replaced immediately by another qualified system. The replacement shall satisfy the same independence requirements. The Nonprofit shall preserve each output, system and model identity, and analysis date. It shall preserve the material inputs and assumptions needed to understand the review.
Section 17.4.3 Urgent Review Exception
Law may require urgent action. Urgent action may also protect people, assets, records, solvency, or organizational continuity.
Despite any contrary numerical requirement, urgent action may proceed with fewer analyses than the applicable tier requires. The Board of Directors must document reasonable efforts to obtain the required number.
The Board of Directors must approve proceeding under applicable voting rules. The authorized human decision-maker remains responsible for every action.
Section 17.5 Governance Reference Assistance
The Nonprofit may operate an AI service that answers questions about the meaning and application of its governing documents, including these Bylaws. The service draws only on the current official version of each document it covers, together with that document’s appendices. It exists so a Director can find a governing provision quickly, not to replace reading it.
Every answer must quote the controlling text and identify the document and provision it came from. Each answer must also identify the version it used. Where a question turns on a defined term, the answer must state that term’s meaning.
Where the governing documents do not address a question, the answer must say so plainly. Where two provisions conflict, the answer must present both and identify the conflict rather than resolve it. Silence and conflict are complete answers, and no answer may supply a rule the governing documents do not contain.
An answer is informational and is not an interpretation of any governing document. The Board of Directors retains its authority to interpret them. No answer authorizes an action, satisfies a required review, or alters a duty imposed by these Bylaws. An answer is not work requiring Independent AI Review.
The Secretary shall receive every question the service reports as unanswered or conflicting. The Secretary shall record those questions and report them to the Board of Directors. That record supports proposals brought under the Amendments Article.
The AI Standards and Protocols shall govern the service’s approved systems, records, and operating requirements.
End of Article XVII. Return to the TOC