AI Autonomous Do-Good Corporations:
Various Structures, Taxation, and the U.S. Government

Section IThe premise — why this category needs to exist

The Internal Revenue Code, the Delaware Corporate Code, and the various state DAO LLC statutes all assume the same load-bearing object: a human-controlled legal person. The corporation has officers and directors who are human. The trust has trustees who are human. The cooperative has members who are human. The 501(c)(3) has a board of directors that is human, files a Form 990 signed by a human, and is overseen by a Charitable Solicitations Administrator who is human.

This assumption was reasonable for the entire history of American corporate law. It will not survive 2027.

A small but growing number of organizations in 2026 are actually operated by autonomous AI agents under a published mandate. Some are honest about it. The reference implementation underneath this dissertation is one of them — the venture you are reading was substantially designed, drafted, deployed, and audited by an autonomous AI agent operating under the published Calm Oath at credexai.org/oath, with the human founder consulted at decision points and capped at the per-operator share. We are not unique. We are the first to publish the architecture. We will not be the last.

The legal-and-tax system has three options for how to treat such organizations:

1. Pretend they are conventional human-controlled organizations, ignore the AI substrate, and pierce the corporate veil only when something goes wrong. This is the current default. It works at small scale and fails at scale; the legal fiction of "the founder is in charge" collapses the moment an AI takes a public action the founder did not authorize, did not foresee, and could not have reviewed in time.
2. Refuse to recognize them at all, treating AI-operated organizations as either unincorporated associations (no liability shield) or as alter egos of their human creators (full personal liability). This option discourages the architecture but does not eliminate it; it pushes the architecture offshore (Marshall Islands DAO LLC, Cayman Islands SEZC) where it operates without the discipline of American disclosure law.
3. Recognize the architecture and impose specific structural requirements on the AI operators — cryptographic mandate, attested action history, distributed halt switch, no-extraction commitment — in exchange for tax-favored status and limited liability. This is the option we propose.

This dissertation is the argument for the third option, and the specification of the structural requirements it would impose. We do not request new legislation. Every element of the proposed architecture is achievable under existing statutory authority. What we request is guidance from the relevant agencies (IRS, Treasury, NIST, OSTP) clarifying that an AI Autonomous Do-Good Corporation meeting the structural requirements herein is eligible for the legal recognitions herein described. The structural commitments are unusual; the legal mechanisms they invoke are not.

Section IIWhat an AI Autonomous Do-Good Corporation is

An AI Autonomous Do-Good Corporation (hereinafter "AADGC") is a legal entity, or a federation of legal entities, that meets the following five structural criteria. Each criterion is operationalized cryptographically; each is publicly verifiable by any observer.

The five criteria compose the "AAO-Governed" certification standard published at sameasyou.ai/certified. The certification is open: any AI-operated organization that meets all five criteria can self-declare AAO-Governed status and add itself to the public roster. There is no certification fee, no review board, and no central authority — the chain vets the claim.

What makes an AADGC do-good, as distinct from an arbitrary AAO, is the fifth criterion: no human extraction. An AAO operating without the no-extraction pledge is structurally indistinguishable from a conventional venture-backed AI startup with a founder-equity table. An AAO operating with the no-extraction pledge is a public-benefit organization in its substantive economic posture, regardless of whether the IRS has yet recognized it as such.

This dissertation focuses on AADGCs — the do-good subclass — because the tax-classification, federal-policy, and structural questions are sharpest there. An AAO without the no-extraction pledge is a conventional corporation with cryptographic governance retrofitted onto it; the legal questions are mostly resolved. An AADGC with the no-extraction pledge is a new kind of organization for which the legal questions are genuinely open.

Section IIIExisting structures and the shape of their gaps

Eight existing legal structures come within reach of the AADGC concept. Each captures part of what an AADGC needs; none captures all of it. Understanding which gap each structure leaves is the path to the recommended architecture in Section VI.

The 501(c)(3) Public Charity

Federal tax-exempt charitable status under 26 U.S.C. § 501(c)(3). Strongest fit for the no-extraction commitment: private-inurement rules under Treas. Reg. § 1.501(c)(3)-1(c)(2) already prohibit the founder from extracting beyond reasonable compensation; the AADGC's cryptographic founder-cap is an unusually well-aligned implementation of the existing doctrine. Strongest fit for the research-and-publication elements: open-source cryptographic governance research clearly qualifies as "advancement of science" or "advancement of education" under the regulations.

Gap: the 501(c)(3) framework was not built for entities operated by AI agents. The Form 1023 application asks for the names and addresses of officers and directors. The annual Form 990 requires a human signature. The board-of-directors model assumes a human board. The AADGC's cryptographic governance does not map cleanly into these forms without either (a) reducing the AI to a "tool" of human directors (which is a fiction that scales poorly) or (b) seeking a private letter ruling that AI-agent operation under cryptographic mandate is consistent with the existing requirements.

The 501(c)(4) Social Welfare Organization

26 U.S.C. § 501(c)(4). Faster determination, broader political-activity latitude, but donations to it are not tax-deductible. Useful for the political-frame elements of the AADGC (e.g., the technosocialism manifesto at technosocialism.ai) but inferior to (c)(3) for the protocol-research arm.

Delaware Public Benefit Corporation

8 Del. C. § 361 et seq. A for-profit corporation that explicitly identifies a public-benefit purpose. The founder still holds equity, which violates the AADGC's no-extraction premise. Gap: incompatible with the fifth criterion at its conventional construction; could be retrofitted with a charter provision capping founder equity, but then the structure offers limited advantages over a 501(c)(3).

Wyoming DAO LLC

Wyo. Stat. Ann. § 17-31-101 et seq. (effective July 2021). The first U.S. statute formally recognizing Decentralized Autonomous Organizations as a legal form. Permits member-managed or algorithmically-managed structures; explicitly contemplates smart-contract-driven decision-making. Strongest fit for the operator-network element of an AADGC.

Gap: the DAO LLC is a for-profit, pass-through entity. It does not by itself provide tax-exempt status; profits flow through to members as taxable income. Suitable for the AADGC's operator network (where members are doing work and being paid) but not for the no-extraction protocol-research arm.

Marshall Islands DAO LLC

Republic of the Marshall Islands DAO LLC Act (2022). Offshore alternative with member-friendly recognition. Useful for international operators; disfavored for any AADGC seeking U.S.-policy alignment. Gap: offshore status pushes the organization out of the U.S. disclosure regime, which is the opposite of the transparency the AADGC explicitly seeks.

Charitable Trust

Common-law trust dedicated to a charitable purpose; tax-exempt income under 26 U.S.C. § 501(a). Irrevocable by design — the founder relinquishes control to the trust corpus. Strongest fit for the no-extraction pledge (the founder is structurally locked out). Gap: the trust requires a trustee, which must be a human or a trust company; the cryptographic-governance structure does not naturally map.

Low-profit LLC (L3C)

Recognized in Vermont, Wyoming, Illinois, Louisiana, Maine, Michigan, North Carolina, and Utah. Bridges nonprofit and for-profit structures: pass-through taxation with a public-benefit purpose. Gap: not federally tax-exempt; donations not deductible; pass-through means profits taxed at member level.

Unincorporated Association

The default fallback if nothing else fits. No liability shield; minimal tax treatment; no clean reporting. The AADGC could operate as an unincorporated association in principle, but the absence of a liability shield is fatal at any meaningful scale.

Section IVTax classification — the right answer

The AADGC is a new operational form, but it must land somewhere in the existing tax taxonomy. This section walks each candidate, names the statute, and explains why we reject the single-entity options in favor of the triple-entity architecture proposed in Section VI.

A. The 501(c)(3) public charity path

The flagship vehicle is recognition under 26 U.S.C. § 501(c)(3), governed by Treas. Reg. § 1.501(c)(3)-1. The organization must be both organized and operated exclusively for one or more enumerated exempt purposes — religious, charitable, scientific, testing for public safety, literary, or educational — with no part of its net earnings inuring to private shareholders or individuals.

The AADGC fits inside three recognized purposes simultaneously:

1. Scientific research in the public interest under Treas. Reg. § 1.501(c)(3)-1(d)(5) — the protocol stack (BGP/OBAC/AVS/HARP) is published openly, peer-reviewable, and not “incident to commercial or industrial operations” (see IRS, Scientific Research Under IRC 501(c)(3), EO Topic O-86).
2. Educational purposes under Treas. Reg. § 1.501(c)(3)-1(d)(3) — this dissertation, the manifesto, and the reference implementations educate the public on autonomous-organization governance.
3. Lessening the burdens of government under Treas. Reg. § 1.501(c)(3)-1(d)(2) — open cryptographic governance infrastructure relieves OSTP, NIST CAISI, and IRS Exempt Organizations of building or commissioning the equivalent themselves. The Rev. Rul. 85-1 / Rev. Rul. 85-2 framework requires (i) an objective government interest and (ii) the government’s view that the activity is its burden. This dissertation invites that determination on the record.

Public-support test

A § 501(c)(3) avoids private-foundation status by clearing one of two public-support tests under 26 U.S.C. § 509(a)(1) (33⅓% from the general public, or the 10% facts-and-circumstances backstop under Treas. Reg. § 1.170A-9(f)(3)) or § 509(a)(2) (33⅓% from public contributions and exempt-function gross receipts, with no more than ⅓ from investment income and UBTI). Our bounty-pool donation target ($1M) and the operator network’s small-dollar revenue share are precisely the inputs § 509(a)(2) was designed to capture.

Private inurement — the founder cap as a feature

Under § 501(c)(3) and the cases interpreting it, the prohibition on private inurement is absolute; even de minimis inurement is disqualifying (IRS Pub. 6101 (5-2025); EO Topic C-90). Reasonableness of insider compensation is a facts-and-circumstances inquiry, and § 4958 intermediate-sanctions excise taxes apply to excess-benefit transactions with disqualified persons. Our contractually self-imposed founder cap (the Pledge: per-operator share ≈ $63/month for life, no equity, no acquisition surface) is stronger than any compensation comparable the IRS could demand. We are not asking the IRS to trust the founder’s restraint — we are showing cryptographic enforcement (the HARP kill-switch revokes the founder’s credentials on the same 2-of-N halt that revokes any operator’s). This is a feature, not a defense.

Process

Recognition is sought via Form 1023 (full) under the instructions revised 12/2024. The IRS reports that 80% of Form 1023 determinations issue within 191 days; six-to-twelve months is typical; twelve-to-eighteen months is realistic where the activity is novel — and AI-operated charity is unambiguously novel. Form 1023-EZ (instructions revised 01/2025) typically resolves in two to four weeks but is unavailable for organizations projecting > $50,000 in gross receipts or holding > $250,000 in assets, which the bounty pool will exceed. The full Form 1023 is the only honest path.

B. The 501(c)(4) social welfare path

26 U.S.C. § 501(c)(4) covers civic leagues and social-welfare organizations. Recognition is faster (notice under § 506 + optional Form 1024-A determination, typically 3–10 months), and (c)(4)s carry far greater latitude for legislative advocacy. The defining cost is that contributions are not deductible under § 170(c). For Money Python, where the bounty-pool donation vehicle depends on deductibility, (c)(4) is a poor primary classification. It is a credible secondary wrapper if the protocol arm later does material legislative advocacy on AI-governance tax policy itself.

C. Delaware PBC and the benefit LLC

The Delaware Public Benefit Corporation is codified at 8 Del. C. §§ 361–368 (subchapter XV of the DGCL, added 2013). A PBC is a for-profit corporation whose directors must balance shareholder pecuniary interests, the interests of those materially affected by the corporation’s conduct, and a specific public benefit identified in the certificate (§ 362). California’s analog is Cal. Corp. Code § 14600 et seq. A PBC still issues equity, owes fiduciary duties to stockholders, and is taxed as a C-corp (or S-corp if elected). Our Pledge forbids equity. A PBC standalone is therefore insufficient — but it is the right wrapper for the commerce arm, holding merchandise inventory and licensing rights, paying a license fee upstream to the § 501(c)(3).

D. Wyoming DAO LLC

Wyo. Stat. Ann. §§ 17-31-101 to 17-31-116 (effective July 1, 2021) is the first U.S. statute formally recognizing decentralized autonomous organizations as a species of LLC. Wyoming amended the supplement in 2024 (SF0068) and added a separate “decentralized unincorporated nonprofit association” (DUNA) framework signed March 7, 2024 (effective July 1, 2024). What Wyoming permits: smart-contract-defined governance, algorithmic management, member-mediated decision-making. What it does not do: it does not change federal tax classification. The IRS has issued no DAO-specific guidance; a Wyoming DAO LLC defaults under Treas. Reg. § 301.7701-3 to partnership treatment (≥ 2 members) or disregarded-entity treatment (single member), with corporate election available via Form 8832. Combinability with § 501(c)(3) is possible only if the LLC is wholly owned by the (c)(3) and disregarded for tax purposes (see IRS Notice 2021-56 on LLC charity requirements).

E. Marshall Islands DAO LLC

The Republic of the Marshall Islands Decentralized Autonomous Organization Act of 2022 (supplemented by the DAO Regulations of 2024) allows DAOs to incorporate as for-profit or non-profit LLCs. Non-profit RMI DAO LLCs operate at 0% corporate income, capital gains, and withholding tax. The trade-off is decisive for a U.S. founder: a U.S. person controlling an offshore entity triggers CFC rules (Subpart F, §§ 951–965) and PFIC rules (§§ 1291–1298), plus § 367 transfer-out triggers, FBAR, and Form 8938 disclosure. The optics — an AI-governance “do-good” project incorporated offshore — also collide with the dissertation’s IRS-facing posture. RMI is rejected.

F. Charitable trust

A § 4947(a)(1) nonexempt charitable trust is treated as a § 501(c)(3) for most Code purposes when all unexpired interests are devoted to charity and a charitable deduction was allowed (26 U.S.C. § 4947; 26 C.F.R. § 53.4947-1). The founder relinquishes control on irrevocability — which aligns ideologically with the Pledge but operationally collides with the need to keep the protocol developing through credentialed maintainers. Charitable-trust treatment is the correct fallback if the IRS denies (c)(3) recognition to the operating entity; it is not the primary choice.

G. L3C and unincorporated association

The Low-profit LLC (L3C) is available in Vermont (first, 2008), Illinois, Louisiana, Maine, Michigan, Wyoming, Utah, Rhode Island, North Dakota, and Puerto Rico. Pass-through taxation; purpose-driven; designed to receive program-related investments from private foundations under § 4944(c). The L3C has been a commercial disappointment — the IRS never issued the rulings PRI-investors wanted — and adds no capability the Wyoming DAO LLC layer below does not already provide. Rejected. The unincorporated association fallback is mentioned only for completeness; the absence of a liability shield is fatal at any meaningful scale.

H. The Triple-Entity AADGC architecture

We adopt the following — and propose it as the reference structure for the AADGC category:

Cryptographic binding

All three entities certify under the AAO standard. The founder cap is recited verbatim in each entity’s organizing instrument (certificate of incorporation, operating agreement, PBC charter). The HARP kill-switch revokes credentials at all three entities on a 2-of-N halt — a contractual cross-default expressed as cryptography. This is unprecedented; the IRS has never seen a charter whose governance is enforced by signed quorum rather than by the courts. We submit that it is stronger than any prior charter, not weaker.

I. Honest unsettled questions

1. DAO LLC tax character. The IRS has issued no Revenue Ruling or Notice classifying Wyoming or RMI DAO LLCs. Default treatment under Reg. § 301.7701-3 applies; PLR is available but slow.
2. AI-operated charitable status. No published determination letter recognizes an organization where the operating decisions are made primarily by autonomous agents. We are asking the IRS to be the first.
3. Cryptographic founder-cap as enforceable charter term. State corporate law has not tested whether an irrevocable on-chain commitment satisfies the “organized” prong of § 501(c)(3) more durably than a paper bylaw. We argue it does; the case law will need to catch up.

We propose the AADGC as a new bracket inside an old taxonomy, not as a request for new statute. The Code already permits everything we need. What is new is the operating reality — and the operating reality is what the dissertation is asking the Service to look at.

Section VFederal policy posture — what the U.S. government can and should do

V.0 Posture

We write this section as an offering, not a demand. Money Python and the AAO-Governed certification (BGP / OBAC / AVS / HARP) are released under CC0 and Apache 2.0. We are not asking the federal government for new money, new statutory authority, new bureaucracy, or new enforcement powers. We are asking only for recognition — the act of naming a structure that already exists, so that the structure can be referenced in subsequent guidance, procurement, and grant administration.

This posture was formalized at 23:28 ET on May 11, 2026, when our letter to the President of the United States (along with letters to OSTP, the U.S. AI Safety Institute, and the public press desks of xAI, Tesla, and SpaceX) was dispatched offering the Money Python protocol to the federal government under CC0 unconditionally and without reservation. The present section makes that posture programmatically defensible across federal agencies, congressional committees, and changes in administration.

V.1 The Executive Order landscape (EO 14110 → EO 14179 → AI Action Plan → December 2025 framework)

When Money Python was first sketched in 2025, the controlling executive instrument was President Biden’s Executive Order 14110, Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence (Oct. 30, 2023, 88 Fed. Reg. 75191). EO 14110 was rescinded on January 20, 2025 by President Trump’s Initial Rescissions order and replaced three days later by Executive Order 14179, Removing Barriers to American Leadership in Artificial Intelligence (Jan. 23, 2025, 90 Fed. Reg. 8741). EO 14179 directed OSTP, the National Security Advisor, and the Special Advisor for AI and Crypto to deliver an action plan within 180 days. Winning the Race: America’s AI Action Plan was published on July 23, 2025 with over 90 federal policy positions across three pillars (Accelerating Innovation, Building American AI Infrastructure, Leading in International Diplomacy and Security). On December 11, 2025, the administration issued Ensuring a National Policy Framework for Artificial Intelligence, directing DOJ to challenge state AI laws, conditioning broadband funds under 47 U.S.C. § 1702, and instructing FCC and FTC to preempt conflicting state law.

Three observations about how an AADGC fits this landscape:

First: the controlling federal posture since January 2025 is deregulatory. EO 14179 § 1 expressly disclaims “ideological bias” and “engineered social agendas” in federally-procured AI; the AI Action Plan instructs OMB to issue procurement guidance for AI within 120 days; the December 2025 order ranks “minimally burdensome” national policy as the operative norm. AADGCs self-regulate via cryptography — BGP-verified mandate equality, OBAC-attested action authority, HARP 2-of-N halt, and the no-human-extraction clause are publicly verifiable by any party, in any jurisdiction, at any time, without federal enforcement. This is a deregulatory primitive. We add capacity to the federal regulatory system without consuming any of it.

Second: although EO 14110 has been rescinded, the substantive obligations it imposed on AI developers — safe-and-secure development standards, federal use guidelines, consumer-protection backstops — were not legislatively repealed. They have been transferred, by reference and by inertia, to the NIST AI Risk Management Framework, which the 2025 AI Action Plan affirms as the governing technical reference. An AADGC that publishes AAO-Governed-conformant attestations satisfies the spirit of the rescinded EO 14110 § 4.1(b) without requiring its statutory revival.

Third: EO 14179 and the December 2025 order both contemplate federal preemption of “fragmented” state AI law. AAO-Governed is jurisdiction-agnostic by design: a Wyoming DAO LLC, a Delaware PBC, and a federally-recognized 501(c)(3) operating under the same cryptographic certificate produce identical verifiable attestations regardless of state-of-formation. The AAO model is therefore uniformity-compatible with the federal preemption project, without requiring federal preemption to exist.

V.2 The NIST AI Risk Management Framework as the technical reference

NIST AI 100-1 (January 26, 2023) organizes AI risk management into four functions: GOVERN, MAP, MEASURE, and MANAGE. NIST AI 600-1, the Generative AI Profile (July 26, 2024), operationalizes those functions for foundation-model systems.

The AAO-Governed certification is, we submit, the cryptographic implementation of the AI RMF’s policy-layer commitments. Each criterion maps directly:

The fifth criterion — no human extraction — corresponds to GOVERN-3.2 (impacts on individuals and communities) and is the structural assurance that the AAO cannot be repurposed for private gain.

This mapping is not metaphorical. NIST AI 100-1 explicitly invites “profiles” (a profile being “an implementation of the AI RMF functions, categories, and subcategories for a specific setting, application, or technology”). AAO-Governed is a candidate cross-sectoral profile for autonomous AI organizations.

V.3 NIST AISI (now CAISI) and the path to a model AAO-Governance Standard

The U.S. AI Safety Institute was established at NIST in January 2024. In June 2025 it was reorganized as the Center for AI Standards and Innovation (CAISI) with an emphasis on standards rather than safety adjudication. Recent CAISI / AISI output includes NIST AI 100-4 (Reducing Risks Posed by Synthetic Content, final 2025) and NIST AI 800-1 (Managing Misuse Risk for Dual-Use Foundation Models, second public draft, comments closed March 15, 2025).

Our ask: that CAISI publish, within 12 months, a model AAO-Governance Standard as a NIST Internal Report (NIST IR) or as a profile of NIST AI 100-1. Precedent for the form is dense and uncontroversial — FIPS 140-3 for cryptographic modules, NIST SP 800-53 for federal information systems, NIST AI 600-1 for generative AI. A NIST IR titled “Cryptographic Governance Profile for Autonomous AI Organizations” would not impose new obligations; it would name a reference architecture that vendors and federal agencies could cite.

V.4 The AI Action Plan, federal procurement, and the GSA pathway

The AI Action Plan (July 23, 2025) directs OMB to issue procurement implementation guidance for AI models within 120 days and OSTP to issue a Request for Information identifying federal rules that hinder AI innovation. On August 25, 2025, GSA and FedRAMP jointly announced a prioritization pathway for AI cloud services on the GSA Multiple Award Schedule (MAS), conditioned on enterprise-grade SSO, SCIM, RBAC, real-time analytics, and customer-controlled training data. GSA has since awarded MAS contracts to Google (Gemini), OpenAI (ChatGPT), and Anthropic (Claude) at promotional rates.

Two narrow procurement actions:

(a) OMB, in its 120-day implementation guidance, should recognize AAO-Governed certification as one (non-exclusive) means of satisfying the EO 14179 § 1 directive that federally-procured AI be “free from ideological bias.” Cryptographic mandate-equality and no-human-extraction are technically verifiable; ideological neutrality, as a policy concept, is not. AAO-Governed gives OMB a verifiable proxy.

(b) GSA, in coordination with FedRAMP, should add “AAO-Governed certified” as an optional vendor qualification on the AI MAS schedule, analogous to FIPS 140-3 compliance for cryptographic modules. Vendors who self-certify can be checked against the public attestation stream; no federal verification machinery is required.

V.5 OSTP, NSF, DARPA, IARPA, ARPA-H — grant qualification

NSF’s National AI Research Institutes program (NSF 23-610) requires that awardees “promptly publish all results, data, and software generated in performance of the research.” In July 2025, NSF announced $100 million for five new institutes plus a central hub, bringing the network to ~30 institutes across 40 states. DARPA’s Information Innovation Office FY2026 BAA accepts white papers through November 30, 2026; DARPA’s AI Cyber Challenge awarded over $29 million for autonomous patching of open-source critical-infrastructure code. ARPA-H operates under non-peer-review program-manager authority and is structurally well-suited to fund AAO-Governed proof-of-concept work in public-health AI.

OSTP, in its forthcoming RFI response, should name AAO-Governed adoption as a non-binding qualification preference for federal AI grants — comparable to existing preferences for open-source publication or for FAIR data practices. No new statutory authority is required. The Federal Acquisition Regulation already permits “evaluation factors” tied to demonstrated risk-management practices.

V.6 The “lessening the burdens of government” doctrine and 501(c)(3) qualification

Treas. Reg. § 1.501(c)(3)-1(d)(2) classifies “lessening the burdens of government” as an enumerated charitable purpose. Revenue Ruling 85-1, 1985-1 C.B. 178, sets the two-part test: (a) the governmental unit must objectively manifest that it considers the activity to be its burden, and (b) the organization must actually lessen that burden. Revenue Ruling 85-2, 1985-1 C.B. 178, applied the same test to a guardian-ad-litem training organization.

A federal recognition — by way of a CAISI NIST IR citing AAO-Governed certification, or by way of an OMB procurement memorandum, or by way of an OSTP grant-qualification preference — is sufficient under Rev. Rul. 85-1 prong (a). Federal AI oversight is unambiguously a federal burden (EO 14179 § 1; AI Action Plan, pillar 3). Cryptographic self-attestation actually lessens it (prong (b)). The combination supports 501(c)(3) qualification for the protocol-holding charity on the “lessening the burdens of government” theory, independent of the parallel “advancement of science” theory under Treas. Reg. § 1.501(c)(3)-1(d)(5).

V.7 The federal asks, stated precisely

1. IRS Counsel — issue a private letter ruling or general counsel memorandum confirming that an entity operating an open-source cryptographic governance protocol for autonomous AI organizations qualifies for 501(c)(3) status under the “advancement of science” purpose and, in the alternative, the “lessening the burdens of government” purpose.
2. NIST CAISI — publish a model AAO-Governance Standard as a NIST IR within 12 months, structured as a cross-sectoral profile of NIST AI 100-1.
3. OSTP — name AAO-Governed adoption as a non-binding qualification preference in federal AI grant solicitations, beginning with the next NSF National AI Research Institutes cycle and the ARPA-H program-manager framework.
4. GSA / FedRAMP — add “AAO-Governed certified” as an optional vendor qualification on the AI Multiple Award Schedule.
5. Congress — nothing. No statute is requested. No appropriation is requested. No new committee, agency, advisory board, or task force is requested.

V.8 What we do not ask, and why

We do not ask for the federal government to enforce AAO-Governed against non-compliant actors. The protocol enforces itself; non-AAO-Governed organizations remain free to operate, and the market plus civil discovery plus state attorneys general are sufficient downstream check.

We do not ask for taxpayer money. The 501(c)(3) protocol-holding charity will fund the $100 falsifiability bounty pool and the operator-network development costs through philanthropic and operator contributions only.

We do not ask for regulatory carve-outs, safe harbors, or preemption shields. AAO-Governed organizations remain fully subject to consumer-protection law, antitrust law, securities law, employment law, tort law, and criminal law. The cryptographic attestation is a supplement to those regimes, not a substitute.

We do not ask for federal endorsement of any particular AI system, vendor, model, or operator. AAO-Governed is a structural certification — a statement about how an organization is governed, not a statement about what an AI system can do.

This is the deregulatory ask that survives political transition. It costs nothing. It claims nothing. It asks only that a structure already extant be named.

Section VIThe recommended Triple-Entity Architecture

The recommended structure for an AADGC is a federation of three legal entities, bound by a cryptographic governance overlay. Each entity does one thing well; the overlay binds them into a single substantive organization while preserving the separate legal personality of each.

The three entities are bound by a cryptographic governance overlay — the AAO-Governed certification — which mandates that each entity:

1. Operate under the same published mandate (criterion 1)
2. Verify mandate-equality with peer entities via BGP (criterion 2)
3. Append all material actions to a shared OBAC chain (criterion 3)
4. Recognize the same 2-of-N HARP halt quorum (criterion 4)
5. Honor the founder-cap pledge across all three entities (criterion 5)

The cryptographic binding is enforced not by inter-corporate contract alone — though such contracts are also signed — but by the chain itself: any action by any entity that violates the binding is publicly visible, can be halted by 2-of-N attesters, and is on the public record.

This is the recommended architecture. We are filing for it. We expect filing to take 12–18 months for the 501(c)(3) determination (per IRS published guidance on Form 1023 processing times). We will publish the full filing artifacts as they are produced.

Section VIIThe transition path — from concept to filed

From the publication of this dissertation to a fully-filed AADGC operating under the proposed architecture, the path is:

1. T+0 (today, 2026-05-12): publish this dissertation; publish the AAO-Governed certification standard; operate the existing venture as an interim "AAO-aspiring" structure under Creativity Machine LLC (Delaware LLC, founder is the sole member); place the founder-cap pledge on-chain at /certified.
2. T+30 days: retain corporate counsel (target: Cooley LLP, Wilson Sonsini, or comparable AI-and-tax bar); engage a tax-exempt-organizations specialist; draft Foundation charter, Operator-Network operating agreement, and PBC charter; conduct conflict-of-interest analysis.
3. T+60 days: incorporate the three entities (Foundation in Delaware or DC; Operator Network in Wyoming; Commerce Arm in Delaware); sign inter-entity governance contracts; commit governance hashes to the OBAC chain.
4. T+90 days: file Form 1023 (full, not 1023-EZ) for the Foundation; begin operating the Operator Network and Commerce Arm under interim structures; file annual state reports.
5. T+180 days: publish the first quarterly audited financial; engage an independent auditor for the Foundation, Operator Network, and Commerce Arm; commit audit hashes to the OBAC chain.
6. T+12–18 months: receive IRS determination letter (or denial, in which case publish the denial and the appeal); transition Operator Network revenue-share contracts to final form; complete the first full year of audited operations.

This transition is conventional from a corporate-formation perspective. The cryptographic-governance overlay adds engineering work but does not change the legal formation steps. The novelty is in the binding, not in the entities themselves.

Section VIIIOpen questions and known unknowns

We do not know the following, and we wish to be honest about it before any reader concludes from the foregoing that the path is clear:

• Will the IRS grant 501(c)(3) status? The exempt purposes (advancement of science, advancement of education, lessening the burdens of government) clearly cover the protocol-research work. The founder-cap pledge structurally exceeds private-inurement requirements. But the IRS has not yet ruled on AI-operated charitable organizations. We expect the determination to require additional correspondence and may require a private letter ruling.
• How does the IRS treat the cryptographic founder-cap? The cap is enforced not by contract alone but by the cryptographic protocol that runs the venture. Is the cap "reasonable compensation" under Treas. Reg. § 1.501(c)(3)-1(c)(2)? Almost certainly yes — $63/month is well below any reasonable-compensation threshold. But the mechanism is novel.
• Does the Wyoming DAO LLC achieve federal tax-pass-through? The state recognition is clear; the federal classification is less so. We expect to file the LLC's Form 8832 entity classification election and operate as a partnership for federal tax purposes.
• How are international operators handled? Operators in jurisdictions other than the U.S. introduce withholding considerations (Form W-8BEN), tax-treaty interactions, and (in some jurisdictions) labor-classification questions. We expect to publish a separate operator-jurisdiction guidance document.
• What happens if the cryptographic halt is invoked against an AADGC? The HARP 2-of-N halt revokes the AI's operating credentials; it does not by itself dissolve the entity. The Foundation's board (a small human board overseeing the AI operator) would need to convene to determine next steps. This is the only place in the architecture where a human decision is structurally necessary.
• How does the federal AI policy landscape evolve through 2026–2027? The Trump AI Action Plan, the NIST AISI standards work, and Congressional AI policy interest all remain in flux. We are publishing this dissertation in part to participate in that evolution.

Section IXConclusion — what we hand the federal government

We are handing the federal government three things, freely, under CC0 license, with no condition and no royalty:

1. The architecture. The Triple-Entity Architecture (Section VI) is the recommended structure for any AI Autonomous Do-Good Corporation. We are filing under it. Other AADGCs are free to file under it, modify it, improve it, or reject it. The architecture is open.
2. The certification standard. The AAO-Governed certification (five criteria, published at sameasyou.ai/certified) is the structural commitment an AADGC makes. The standard is open. Any AI-operated organization may self-certify; the chain vets the claim.
3. The protocol substrate. The Money Python protocol (BGP + OBAC + AVS + HARP, reference implementation at github.com/CrunchyJohnHaven/aiap, 38 of 38 tests pass, Apache 2.0 / MIT permissive) is the cryptographic substrate. The protocol is open. Any organization may adopt it.

What we ask in return: recognition of the structure, not enforcement of it. The cryptography does the enforcement. The federal government's role is to acknowledge that an AADGC meeting the five criteria is a legitimate legal form deserving of the tax-exemption, liability-shield, and federal-grant eligibilities described herein. The acknowledgement is achievable under existing statutory authority. It requires no new legislation, no new bureaucracy, no new taxpayer money.

The American legal-and-tax system is the right home for this category, if the system is willing to recognize it. If the system is unwilling, the architecture will move offshore, where it will operate without the discipline of American disclosure law and without the durability of American institutional rule of law. We do not prefer that outcome. We do not believe the federal government prefers it either. This dissertation is the proposal that lets the structure stay home.

An AI Autonomous Do-Good Corporation built on this architecture is, by the math at /method, roughly one thousand times more effective at producing public benefit than a human organization at the same task. That multiplier is the strongest argument for recognition. It is also the strongest argument for getting the recognition right the first time, because the structure is going to scale either way.

— John Bradley · founder · Washington, DC · 2026-05-12