The conversion.
There is an established way to turn a charity into a company, and OpenAI did not use it. The standard mechanism — proven in the 1990s, when California oversaw a wave of nonprofit-to-for-profit conversions in healthcare — is divestiture: the charity sells its assets at independently appraised fair market value, the proceeds endow a fully independent foundation with a similar mission, and the charity exits the for-profit entirely. When Blue Cross of California converted, it funded two independent foundations with cash and stock worth over $3 billion. When Health Net converted, it created the California Wellness Foundation. The charity got its value out, an independent steward got the proceeds, and the new for-profit went its own way.
OpenAI did something structurally different. The nonprofit — now the OpenAI Foundation — did not sell its assets and exit. It kept control of the for-profit, holding roughly $130 billion in equity rather than cash, and continues to govern the OpenAI Group PBC it converted into. The charity did not get its value out and hand it to an independent steward. It kept the value, kept the control, and kept the entanglement.
That difference is not a technicality. It goes to the three tripwires that charitable-asset law is built around. Charitable assets are subject to an “asset lock” — the charitable trust doctrine holds they are permanently dedicated to the nonprofit’s purpose and cannot be distributed to private hands; the private-inurement rule forbids any of the charity’s value flowing to private individuals; and the fair-market-value rule requires the charity receive full value for anything it transfers. The healthcare playbook cleared all three by divesting at appraised value into an independent foundation. OpenAI cleared them by keeping control — which is a far less tested path.
And the people whose job is to police those tripwires let it through. California’s Attorney General Bonta and Delaware’s Kathy Jennings, after nearly a year of investigation, blessed the recapitalization on October 28, 2025 — on the representation that nonprofit control is preserved — despite having the standing, resources, and jurisdiction to challenge it. Critics had called the nonprofit “little more than a rubber stamp of the for-profit.” The AGs settled anyway. The question that settlement leaves open is not about OpenAI specifically. It is about every charity that follows.
The structural argument I want to make: OpenAI’s conversion did not follow the established divestiture playbook — in which a charity sells its assets at fair value and endows an independent foundation — but instead used a control-retention model, in which the nonprofit keeps its equity and its control of the for-profit, and the gap between those two models is the precedent the conversion sets: that charitable assets can migrate into private-equity-style structures as long as some equity stake flows back to the nonprofit and the nonprofit nominally retains control, which is a far weaker protection than divestiture and which the attorneys general blessed rather than tested. This is the fifth AI Governance dispatch and the tax-and-charitable-law companion to The prospectus: where the prospectus asked what the conversion must disclose to investors, this asks what the conversion did to charitable-asset law.
The headline integrative finding: The honest both-sides read is that the control-retention model is either a genuine innovation that protects the mission better than divestiture, or a loophole that guts charitable-asset law — and which one it is depends entirely on whether nonprofit control is real or nominal. The charitable case for control retention is real: a foundation that keeps a $130 billion stake and governs the for-profit stays in the game, with resources and influence a cash-out foundation would lack, and its mission (ensuring AGI benefits humanity) is arguably better served by steering the company than by funding grants from the sidelines. The charitable case against is equally real: a nonprofit that controls the for-profit it depends on for its value, staffed and influenced by the for-profit’s interests, is structurally compromised in exactly the way the asset-lock and inurement rules were built to prevent. The deepest point is that the entire protection now rests on a single contested fact — whether the OpenAI Foundation actually controls OpenAI Group or merely appears to — and that fact is precisely the thing that cannot be verified in advance, only observed when the two interests conflict. The AGs blessed the structure on the paper version of control. Whether the real version holds is the experiment now running, and the precedent it sets for the next decade of conversions depends on the answer.
This essay walks the established divestiture playbook, the control-retention model OpenAI used instead, the three tax-law tripwires, the valuation problem, the attorneys general who blessed rather than tested, the precedent for every charity that follows, and the structural reading of a conversion that redefined what “nonprofit” can become.
The conversion.
What turning the largest
nonprofit into a company
did to charity law.
held, not divested for cash
independent foundations (Blue Cross)
that nonprofit control is preserved
set by settlement, not adjudication
- Charity sells assets at appraised fair value
- An independent foundation inherits the proceeds (Blue Cross → $3B+)
- The charity exits the for-profit entirely
- Protection = the value leaves the for-profit’s control
- Foundation keeps ~$130B equity, not cash
- Keeps controlling the OpenAI Group PBC
- No exit — the value stays inside the company
- Protection = nominal nonprofit control of the for-profit
The conversion redefined what a nonprofit can become — and did so by acquiescence rather than adjudication, on a representation the enforcers accepted rather than a standard a court imposed. The experiment is now running, and the next decade of conversions is watching the result.Thorsten Meyer · The Conversion · AI Governance 05
By Thorsten Meyer — June 2026
This is the fifth dispatch in the AI Governance track — the corporate-structure forensics of the labs. The first four walked the Musk verdict, the structural mirror, the AGI clause, and the prospectus. This one walks the charitable-law mechanics underneath the prospectus: not what the conversion must disclose, but what the conversion did to the centuries-old rules governing charitable assets — and the precedent that did to American charity law.
The structural argument I want to make: charitable-asset law has, for centuries, rested on a simple promise — that assets given to a charity stay dedicated to charity, permanently — and the control-retention conversion tests whether that promise survives contact with an asset valuable enough to make breaking it worth the legal effort. The healthcare conversions honored the promise by divesting: the value left the charity’s control and went to an independent steward. The OpenAI conversion keeps the value inside a structure the for-profit influences, and asks the law to accept that nominal nonprofit control is protection enough. Whether it is, is the whole question.
The headline integrative finding: The conversion sets a precedent that will outlive the specifics: charitable assets can be migrated into for-profit structures without divestiture, as long as the nonprofit keeps equity and nominal control. If that precedent holds, the asset lock becomes a turnstile — a charity is a tax-advantaged staging ground for whatever later proves lucrative, and “nonprofit” means whatever the founders decide once the asset gets valuable. If it does not hold — if the foundation’s control proves real and the mission genuinely steers the company — then control retention is a legitimate modern alternative to divestiture, and the critics were wrong. The precedent is set; its meaning is not. And because it turns on whether nominal control becomes real control, it will be settled not by the settlement documents but by what happens the first time the Foundation’s mission and the company’s profit genuinely diverge.
This essay walks the divestiture playbook (Section I), the control-retention model (Section II), the three tripwires (Section III), the valuation problem (Section IV), the attorneys general (Section V), the precedent (Section VI), and the structural reading (Section VII).
Managing Modern Healthcare: Knowledge, Networks and Practice (Routledge Studies in Health Management)
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
I · The established playbook · how charities became companies before
The precedent crystallization. There is a proven, decades-old mechanism for converting a charity into a for-profit, and it works precisely because it severs the charity from the company. Understanding the old way is the key to seeing what is new about OpenAI’s.
The healthcare conversions
Divestiture into an independent foundation: in the 1990s, California oversaw multiple nonprofit-to-for-profit conversions in healthcare, and they followed a consistent pattern. Blue Cross of California transferred its managed-care business to its for-profit subsidiary, WellPoint, and — under regulatory pressure — funded two independent foundations (The California Endowment and The California HealthCare Foundation) with cash and stock worth over $3 billion. Health Net’s conversion created the California Wellness Foundation. The charitable value was extracted, measured, and handed to an independent steward whose only job was the mission.
Why divestiture is the protective mechanism
Severance is the protection: the reason divestiture protects charitable assets is that it separates them from the for-profit’s influence. The new foundation holds the proceeds, governs itself, and pursues the mission free of the converted company’s commercial interests. The charity gets full value and exits; the foundation that inherits the value answers to no one but the mission. Even the stock-based conversions, which kept a temporary link, were structured with an explicit goal of full divestiture over time — because the law recognizes that a foundation holding its “donor’s” stock is an unhealthy marriage of interests until the stock is sold.
The principle underneath
The charitable value must leave the for-profit’s control: the deep principle is that charitable assets, once dedicated, cannot remain under the control of the people or entity that would profit from them. Divestiture enforces this by design — the value moves out, to an independent body. The old playbook protects charitable assets by getting them away from the for-profit, not by keeping them entangled with it.
The playbook observation
The established mechanism for converting a charity into a company is divestiture: the charity sells its assets at appraised fair value, an independent foundation inherits the proceeds, and the charity exits the for-profit entirely — proven in the 1990s healthcare conversions (Blue Cross/WellPoint funding $3B+ into independent foundations, Health Net into the California Wellness Foundation). Divestiture protects charitable assets by severing them from the for-profit’s control. This is the baseline against which OpenAI’s conversion must be measured — because OpenAI did not divest, and the entire question is whether the alternative it used protects charitable assets as well as severance does.
The Nonprofit Board of Trustees Record Book: Official Minutes, Resolutions, and Corporate Governance Log
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
II · The control-retention model · what OpenAI did instead
The departure crystallization. OpenAI’s conversion inverts the protective logic of divestiture. Instead of getting the charitable value away from the for-profit, it keeps the charity inside the for-profit’s structure — as its controlling parent. That inversion is the innovation, and the risk.
The structure
The Foundation keeps equity and control: under the October 28, 2025 structure, the nonprofit became the OpenAI Foundation, the for-profit became OpenAI Group PBC, and the Foundation holds roughly $130 billion in equity while continuing to control the Group. There was no divestiture, no independent foundation funded with cash, no exit. The charity kept its value inside the company — as equity, not proceeds — and kept governing the company it converted.
Why this inverts the protection
Entanglement instead of severance: where divestiture protects by separating the charity from the for-profit, control retention does the opposite — it binds them. The Foundation’s $130 billion of value is the for-profit; its mission depends on the for-profit’s success; and its governance is exercised over the very entity whose commercial interests the charitable-asset rules were designed to wall off. The structure that was supposed to protect the charity from the for-profit’s influence now makes the charity the for-profit’s parent — which means the protection runs through the entanglement the old rules forbade.
The charitable case for it
Steering beats funding from the sidelines: there is a real argument for the model. A foundation that keeps a $130 billion stake and controls the company has resources and influence a cash-out foundation could never match, and OpenAI’s mission — ensuring AGI benefits humanity — is plausibly served better by steering the company building AGI than by funding grants from outside it. Control retention can be read as a mission-protective innovation: keep the charity in the game, at the helm of the thing that matters, rather than cashing out and watching from the sidelines. Whether that reading holds depends on whether the control is real.
The control-retention observation
OpenAI’s conversion used a control-retention model that inverts divestiture’s protective logic: instead of severing the charity from the for-profit, it keeps the Foundation holding $130 billion in equity and controlling the OpenAI Group PBC — binding the charity to the company rather than freeing it from it. There is a genuine charitable case for this (steering the AGI company beats funding grants from the sidelines), and a genuine charitable risk (the charity is now entangled with the for-profit in exactly the way the old rules forbade). The model’s legitimacy turns entirely on whether the Foundation’s control is real or nominal — which is the question the next sections, and the next decade, have to answer.
Good Counsel: Meeting the Legal Needs of Nonprofits
Used Book in Good Condition
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
III · The three tripwires · the tax-law rules the conversion had to clear
The legal-mechanics crystallization. Charitable-asset law is built around three rules, and any conversion has to clear all three. The healthcare playbook cleared them by divesting. OpenAI cleared them by other means — and how it cleared each one is where the precedent lives.
The asset lock
The charitable trust doctrine: a 501(c)(3) nonprofit’s assets are permanently dedicated to its charitable purpose and held in trust for the public; they cannot be distributed to private hands. This “asset lock” is why a direct conversion is impossible — the assets cannot simply become private equity. OpenAI cleared it by keeping the assets nominally inside the charitable structure (the Foundation still holds them) while making them economically operative inside the for-profit (as equity in the PBC) — neither fully divested nor fully locked, but a hybrid the doctrine was not written to address.
Private inurement
No charitable value to private hands: the private-inurement rule forbids any of the charity’s net earnings or value benefiting private individuals — founders, directors, insiders. Conversions risk this when equity, compensation, or control flows to “disqualified persons.” The control-retention structure raises the question without resolving it: when the Foundation controls a for-profit whose equity holders include employees and investors, and whose leadership overlaps, does charitable value flow to private hands? The structure is designed to say no — the Foundation holds the controlling value — but the entanglement is exactly the condition inurement rules were built to scrutinize.
Fair market value
Full value for what is transferred: the charity must receive at least fair market value for any assets it transfers to the for-profit, to avoid private benefit. In a divestiture, this is enforced by an independent appraisal and a cash sale. In a control-retention conversion, “fair value” is murkier — the Foundation did not sell at an appraised price; it received equity whose value (the ~$130 billion figure) is a function of the company’s own valuation, which the Foundation, as controlling parent, is entangled with. The fair-value test, designed for an arm’s-length sale, strains when the buyer and seller are the same governance structure.
The tripwires observation
The conversion had to clear three charitable-law tripwires — the asset lock (assets permanently dedicated, undistributable), private inurement (no value to private hands), and fair market value (full value for transfers) — and where divestiture clears them by an arm’s-length sale into an independent foundation, control retention clears them by a hybrid the rules were not written to address: assets nominally locked but economically operative in the for-profit, value held rather than sold, control retained rather than severed. Each tripwire is technically cleared and substantively strained. The conversion did not break the rules; it found the space between them — and that space is the precedent.
LoneStar Tracking Hidden GPS Tracker Device Yabby3 – 4G/5G Weatherproof Car, Trailer, Vehicle & Asset Tracking, Anti-Theft Satellite Tracking, Real-Time Location & Security(Low Cost Subscription Plan)
HIDDEN CAR TRACKER DEVICE WITH EXCEPTIONAL BATTERY LONGEVITY: Our Vehicle GPS Tracker, powered by a mere 3 AAA…
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
IV · The valuation problem · what is $130 billion of a mission worth?
The measurement crystallization. Even setting aside the structural questions, the conversion rests on a number — roughly $130 billion — and how that number was set, and what it represents, is its own forensic problem. Valuation is, as the healthcare conversions found, the most controversial step.
The contested figure
$130 billion of equity, not cash: the Foundation’s stake is reported at approximately $130 billion — but that is the value of equity in a private company, set by the company’s own funding-round valuations, not an independent appraisal of charitable assets sold at arm’s length. Earlier estimates of the charitable assets at stake ran as high as $157 billion. The number is large and the number is soft: it is a mark on private equity, not a price in a market sale, and it moves with the company’s valuation rather than reflecting an independent measure of what the public is owed.
Why valuation is the controversial step
The public’s interest rides on the number: in a conversion, the valuation determines how much charitable value is preserved — too low, and the public is shortchanged; the foundation that inherits it is undersized relative to what the charity was worth. The healthcare conversions fought hardest over valuation precisely because it sets the size of the public’s continuing benefit. A conversion that gets the valuation wrong transfers public value to private hands by understatement — which is why independent appraisal is the protective norm, and why equity-held-rather-than-cash-sold is a weaker protection.
The entanglement compounds it
The controlling parent values its own stake: in the control-retention model, the Foundation’s stake is valued by reference to a company the Foundation controls — so the entity whose interest is in a high valuation (to maximize charitable assets) is entangled with the entity whose past valuations set the number. There is no arm’s-length seller and buyer; there is one governance structure on both sides of the valuation, which is exactly the conflict the fair-value rule exists to prevent.
The valuation observation
The conversion rests on a soft, contested number — roughly $130 billion of private-company equity, not an independently appraised arm’s-length sale price — set by reference to a company the Foundation controls, which collapses the buyer-seller separation the fair-value rule depends on. Valuation is the most controversial step in any conversion because the public’s continuing benefit rides on it, and equity-held-rather-than-cash-sold is a weaker protection than an independent appraisal. The $130 billion figure is simultaneously enormous and unverified as a measure of what the public is owed — which is a problem the divestiture playbook solves with an arm’s-length sale and the control-retention model leaves open.
V · The attorneys general · who blessed rather than tested
The enforcement crystallization. Charitable-asset law has a designated enforcer: the state attorney general, who represents the public as the ultimate beneficiary of charitable assets. Two of them — California’s and Delaware’s — had this conversion in front of them, and what they did with it is central to the precedent.
The enforcers’ role
The AG represents the public’s interest: the attorney general monitors charitable conversions on common-law and statutory authority, ensuring the charity receives fair value and the public’s interest is protected; in disputes, the AG sues in court as the public’s representative. California’s Bonta and Delaware’s Jennings were the designated guardians of the charitable assets at stake — the parties with standing, resources, and jurisdiction to challenge the conversion if it shortchanged the public.
What they did
They settled, on a representation: after nearly a year of investigation — California had cited OpenAI’s articles dedicating its assets “irrevocably” to charitable purpose, and its own “responsibility to protect assets held in charitable trust” — both AGs blessed the recapitalization on October 28, 2025. Bonta issued a settlement statement; Jennings released a Statement of No Objection tracking California’s almost word for word. The blessing rested on the representation that nonprofit control is preserved — the paper version of control, not a tested one.
Why the settlement is the issue
They had standing and chose not to use it: the deeper question, as critics framed it, is why the case settled at all. Both offices had the authority to litigate the core question — whether a charity funded by tax-deductible donations can be converted into a corporation. They chose not to. A test case with the standing, resources, and jurisdiction to set the law was resolved by settlement instead — which means the precedent was set by acquiescence, not adjudication, and the hardest question (is nominal control real control?) was never put to a judge.
The attorneys-general observation
The designated enforcers of charitable-asset law — California’s and Delaware’s attorneys general — blessed the conversion on October 28, 2025 on the representation that nonprofit control is preserved, despite having the standing, resources, and jurisdiction to litigate the core question, and despite critics calling the nonprofit “little more than a rubber stamp of the for-profit.” The precedent was set by settlement, not adjudication. The hardest question — whether nominal nonprofit control is real protection — was the one the enforcers were positioned to test and chose not to, which means the protection now rests on a representation the guardians accepted rather than a standard a court imposed.
VI · The precedent · what this does to every charity that follows
The generalization crystallization. The conversion’s largest consequence is not what it did to OpenAI but what it does to the next charity. A precedent set by the most valuable nonprofit-to-for-profit conversion in history will shape every one that follows.
The precedent stated
Charitable assets can migrate if equity flows back: the precedent the conversion sets is that charitable assets can be migrated into private-equity-style structures without divestiture, as long as some equity stake flows back to the nonprofit and the nonprofit nominally retains control. The asset lock, on this precedent, is not a wall but a turnstile: a charity can become a company, keep the founders and investors in the equity, and satisfy the law by keeping the nonprofit as a controlling shareholder rather than divesting it.
Why the precedent is dangerous
The staging-ground problem: if the precedent holds, a nonprofit becomes a tax-advantaged staging ground for whatever later proves lucrative. Found a charity, take tax-deductible donations, build something valuable under charitable protection, and — once the asset is valuable enough to make conversion worth it — migrate it into a for-profit while keeping the charity as a controlling shell. “Nonprofit” would mean whatever the founders decide once the asset gets valuable, and the centuries-old promise that charitable assets stay charitable would become a temporary state that ends when the money gets big enough. That is the warning from people who run nonprofits that have stayed nonprofits.
Why the precedent might be benign
Control retention might genuinely protect the mission: the other reading is that the precedent is fine — that control retention, done honestly, keeps the charity at the helm of the most important asset with more resources and influence than divestiture would give it, and that a $130 billion foundation steering the AGI company is a better outcome for the mission than a cash-out foundation funding grants. If nonprofit control proves real, the precedent is not a loophole but a modernization — a recognition that for some missions, steering beats severance. The precedent is dangerous or benign depending entirely on whether control is real.
The precedent observation
The conversion sets a precedent that will outlive its specifics: charitable assets can migrate into for-profit structures without divestiture, as long as equity flows back and the nonprofit nominally retains control — which is either a loophole that turns the asset lock into a turnstile and makes “nonprofit” a temporary status, or a modernization that keeps charities steering their most valuable assets, depending entirely on whether the retained control is real. A precedent set by the largest such conversion in history will shape the next decade of them. And because the precedent turns on the reality of control, it cannot be evaluated from the settlement documents — only from what happens when mission and profit first genuinely conflict, which is the experiment now running across the entire nonprofit sector.
What this is not
It is not a claim that OpenAI broke the law. The conversion cleared the legal process; the AGs blessed it. The claim is that it cleared the tripwires by finding the space between them, not by the divestiture that previously defined compliance.
It is not a claim that control retention is necessarily worse than divestiture. It may protect the mission better, if control is real. The claim is that its protection is untested and rests on a contested fact.
It is not a prediction that the precedent will be abused. It may not be. The claim is that the precedent now exists, that it weakens the asset lock, and that whether it is abused depends on enforcement that just declined to test it.
The synthesis observation
OpenAI’s conversion used a control-retention model rather than the established divestiture playbook, cleared the three charitable-law tripwires by finding the space between them rather than by an arm’s-length sale into an independent foundation, rested on a soft $130 billion valuation set by reference to a company the Foundation controls, and was blessed by attorneys general who had the standing to test the core question and chose to settle instead — setting the precedent that charitable assets can migrate into for-profit structures as long as equity flows back and nominal control is retained. Whether that precedent is a loophole or a modernization turns on the single contested fact of whether the Foundation’s control is real.
There is no single answer. Anyone offering one is selling something. What is unambiguous is that the conversion redefined what a nonprofit can become — and did so by acquiescence rather than adjudication, on a representation the enforcers accepted rather than a standard a court imposed. The asset lock that for centuries kept charitable assets charitable is now, by precedent, a structure a sufficiently valuable charity can convert through while keeping its founders and investors in the equity and its nonprofit as a controlling shell. If the control proves real, that is a modernization. If it proves nominal, it is the moment “nonprofit” started meaning whatever the founders decide once the asset gets valuable enough — and the guardians who could have tested which it would be chose not to. The experiment is now running, and the next decade of conversions is watching the result.
That is the structural editorial question the conversion sits on top of. It is a departure from the divestiture playbook that protected charitable assets by severing them. It is three tripwires cleared by finding the space between them. And it is a precedent set by settlement, turning on a contested fact about control that only conflict will reveal. And it is the charitable-law foundation under everything the prospectus disclosed — because before the conversion could become an S-1 risk factor, it had to become a legal fact, and the legal fact is that American charity law now permits a charity to become a company while keeping its value inside it, blessed by the guardians who declined to test whether the keeping was real.
About the Author
Thorsten Meyer is a Munich-based futurist, post-labor economist, and recipient of OpenAI’s 10 Billion Token Award. He spent two decades managing €1B+ portfolios in enterprise ICT before deciding that writing about the transition was more useful than managing quarterly slides through it. He runs StrongMocha News Group, a network of more than 450 niche WordPress magazines built on the DojoClaw editorial engine. More at ThorstenMeyerAI.com.
Related Reading · the AI Governance track
This dispatch
- This piece · The conversion · the charitable-law forensic — how OpenAI’s conversion departed from the divestiture playbook, cleared the three tax-law tripwires by control retention rather than severance, and set the precedent that charitable assets can migrate into for-profit structures as long as equity flows back · empirical-clay dominant, structural-slate and labor-rose balance
The track
- The calendar technicality · AI Governance 01 · the Musk verdict on the conversion this piece’s charitable-law analysis underpins
- The structural mirror · AI Governance 02 · the two labs’ governance structures
- The clause · AI Governance 03 · the AGI clause inside the converted structure
- The prospectus · AI Governance 04 · the securities-disclosure companion — the prospectus asks what the conversion must disclose; this asks what the conversion did to charitable-asset law
Adjacent tracks
- The stake · Post-Labor 01 · who owns the converted equity, and the broader question of broad-based ownership the conversion’s concentration sharpens
- The runway · Enterprise Reorg 04 · the valuation pressure (the funding-round clawbacks) that drove the conversion’s aggressive deadline
Sources
The established playbook
- San Francisco Foundation · Coalition requests AG action — the 1990s healthcare precedent: Blue Cross of California transferred its managed-care business to WellPoint (1993) and funded The California Endowment and The California HealthCare Foundation with cash and stock worth over $3 billion; Health Net created the California Wellness Foundation; the charitable assets at stake in OpenAI estimated as high as $157 billion · sff.org
- Health Affairs · Managing charitable assets from conversions — stock-based conversions force “a potentially unhealthy marriage of interests between the charity and its for-profit donor until the stock can be prudently divested”; IRS private-foundation restrictions on self-dealing and excess holdings; the explicit goal of full divestiture over a reasonable period · healthaffairs.org
- The Common Law Power of the Legislature (PMC) — the valuation as “often the most controversial step”; assets equal to fair market value transferred to a new charitable organization with a substantially similar purpose; the AG monitors on common-law and statutory authority and sues as the public’s representative when there are disputes · pmc.ncbi.nlm.nih.gov
The control-retention model
- OpenAI · Our structure / Built to benefit everyone — the October 28, 2025 structure: the nonprofit is now the OpenAI Foundation, the for-profit is OpenAI Group PBC; “the OpenAI Foundation continues to control the OpenAI Group”; the Foundation holds equity “currently valued at approximately $130 billion,” with additional ownership granted as the for-profit reaches a valuation milestone · openai.com
- EA Forum · Inside OpenAI’s controversial plan — the deadline pressure: failure to convert by end-2025 would have required returning $20B of the $40B round (at a $300B valuation); the $6.6B October investment required conversion by October 2026 to avoid clawbacks with 10% interest; Chan Loui — “a very aggressive deadline” regulators may struggle to accommodate · forum.effectivealtruism.org
The three tripwires
- LegalClarity · Can a nonprofit convert to a for-profit? — the charitable trust doctrine / “asset lock”: all 501(c)(3) assets permanently dedicated to charitable purpose, held in trust, undistributable to private individuals (the private-inurement prohibition); the asset sale at fair market value; the independent third-party appraisal submitted to the AG; the IRS notification of status change · legalclarity.org
- Harvard Corp Gov / Key legal considerations in nonprofit spinouts — fair value: the nonprofit must receive at least fair market value to avoid private inurement and private benefit; establishing the for-profit as a PBC with a mission aligned to the charitable purpose to mitigate public reaction; the disqualified-persons compensation rules (5-year lookback) · corpgov.law.harvard.edu
- Caplin & Drysdale · Conversions of nonprofit organizations — restrictions that come with tax-exempt status continue to apply even after relinquishing it; the assets must be dedicated to an exempt purpose; the conversion must not result in private benefit or inurement; these IRS concerns are in addition to state charity-regulator review · caplindrysdale.com
- SE4Nonprofits · Q&A 173 — “converting” a 501(c)(3) into a taxable for-profit is rarely available under state nonprofit law; the excess-benefit, inurement, and private-benefit pitfalls; the organizational test requiring assets be permanently dedicated · se4nonprofits.com
The attorneys general and the precedent
- Route Fifty · California is investigating OpenAI’s conversion — the December 6 letter from deputy AG Christopher Lamerdin citing OpenAI’s articles (“OpenAI’s assets are irrevocably dedicated to its charitable purpose”) and the office’s “responsibility to protect assets held in charitable trust”; the request for information on asset transfers and valuation · route-fifty.com
- GeekWire (Oren Etzioni op-ed) · Don’t let the OpenAI soap opera hide the precedent — Public Citizen’s September 2025 letter: the nonprofit has become “little more than a rubber stamp of the for-profit”; Bonta’s October 28 settlement statement and Jennings’s same-day Statement of No Objection “tracking California almost word for word”; the AGs had standing, resources, and jurisdiction and chose not to litigate; the precedent that “charitable assets can be migrated into private-equity-style structures as long as some equity stake flows back” · geekwire.com
- CalMatters · OpenAI’s restructuring deal is full of holes — the deal addresses Bonta’s concerns but could “revive worries that OpenAI is misusing charitable tax exemptions”; the nonprofit nominally in control but “numerous ways the for-profit company could end up calling the shots” · calmatters.org
The governance-track backbone
- The prospectus · Thorsten Meyer · AI Governance 04 · the securities-disclosure companion to this charitable-law forensic — the conversion this piece analyzes is the heaviest item in the S-1 the prospectus dispatch examined
Key reference figures crystallized
- The divestiture playbook: Blue Cross/WellPoint (1993) → The California Endowment + The California HealthCare Foundation, $3B+; Health Net → California Wellness Foundation; charity sells at appraised value, independent foundation inherits, charity exits; “full divestiture over a reasonable period” the norm even for stock conversions
- The control-retention model: Oct 28, 2025 — nonprofit becomes OpenAI Foundation, for-profit becomes OpenAI Group PBC; Foundation holds ~$130B equity (earlier estimates to $157B) and “continues to control” the Group; no divestiture, no independent cash-funded foundation, no exit; entanglement instead of severance
- The three tripwires: the asset lock / charitable trust doctrine (assets permanently dedicated, undistributable); private inurement (no value to private hands); fair market value (full value for transfers) — cleared by control retention “in the space between them” rather than by arm’s-length sale
- The valuation problem: ~$130B is private-company equity (a mark, not a market price), set by reference to a company the Foundation controls; valuation the most controversial step because the public’s continuing benefit rides on it
- The attorneys general: Bonta (CA) + Jennings (DE) blessed Oct 28, 2025 on the representation that nonprofit control is preserved; the Dec 6 investigation letter (assets “irrevocably dedicated”); Public Citizen “rubber stamp”; settled despite standing, resources, jurisdiction — precedent by acquiescence, not adjudication
- The deadline driver: end-2025 conversion or return $20B of the $40B round; Oct 2026 deadline on the $6.6B October investment to avoid clawbacks at 10% interest — the valuation pressure behind the “aggressive deadline”
- The precedent: charitable assets can migrate into for-profit structures without divestiture if equity flows back and nominal control is retained; the asset lock as turnstile vs modernization; turns on whether control is real, revealed only when mission and profit conflict
