A decentralised autonomous organisation —a DAO— is an elegant idea taken to the extreme. Instead of a company with a board, a headquarters and a chief executive, it proposes a structure that lives on a blockchain: decisions are made by votes of those who hold the governance tokens, the rules execute themselves through smart contracts, and there is, in theory, no one “in charge”. The movement’s founding motto summed it up with ambition: the code is law. If everything is written in the smart contract and runs automatically, why would the old law be needed?
The courts are providing the answer, and it is not the one the movement expected. In a series of decisions that culminated in 2025, US judges have concluded something uncomfortable for the decentralised ideal: when a DAO causes harm or breaches an obligation, someone has to answer, and absent a recognised legal person, that someone can be the individuals who hold the tokens and vote. The code, it turns out, was not the law. It was, at most, a set of instructions operating within a world where the law still rules.
This analysis traces how that gap opened between the ideal of the ownerless organisation and the legal reality of liability, and how different jurisdictions are trying to close it.
The underlying problem: an entity the law does not contemplate
Classic commercial law knows corporations, limited-liability companies, associations, foundations, cooperatives. Each has one decisive trait: legal personhood, the legal fiction by which an organisation is “someone” distinct from its members, able to own assets, sign contracts, sue and be sued, and —crucially— to limit its members’ liability to what they contributed. If a limited company goes bankrupt, its partners lose their investment, but not their home.
A DAO, as conceived, lacks that personhood. It is not registered in any commercial registry, it has no registered office, and its members may be spread around the world, connected only by holding a token. That absence, which for the decentralised ideal was a virtue —no one controls, no one captures— is for the law a void to be filled. And the way courts have filled it is what set off the alarms: absent a better figure, they have treated DAOs as general partnerships or unincorporated associations, the two oldest and most dangerous organisational forms in law, because neither offers limited liability. In a general partnership, each partner answers with all their assets for the debts of the whole.
The chain of rulings that closed the ideal
Three US cases, chained together, mark the trajectory.
| Case | Year | What the court decided |
|---|---|---|
| Sarcuni v. bZx DAO | 2023 | Refused to dismiss the suit: a DAO “is possibly a general partnership”, and its token holders could answer personally |
| CFTC v. Ooki DAO | 2023 | First ruling to treat a DAO as a “person” and unincorporated association subject to liability as an entity |
| Samuels v. Lido DAO | 2025 | A California federal court opened the door for governance-token holders to answer as de facto partners |
The bZx case was the first warning. After a security breach, a class action argued that the governance-token holders were liable. The judge not only declined to dismiss the case, but paid special attention to the founders’ attempts to evade US laws by transferring ownership from limited companies to a DAO, citing the founders’ own words about that manoeuvre. The implicit message was that disguising a company as a DAO does not erase a company’s responsibilities.
The Ooki DAO case took the logic further. The CFTC —the US derivatives regulator— obtained a default judgment in which the court accepted that the DAO was a “person” and an unincorporated association under federal and state laws. The director of the CFTC’s enforcement division celebrated the decision as “precedent-setting” and “a wake-up call to anyone who believes they can circumvent the law by adopting a DAO structure”.
But it was the Lido DAO ruling, in 2025, that truly shook the sector. By treating governance-token holders as partners in a general partnership, the California court opened the possibility that participating in governance —something understood as a near-consequence-free technical activity— carries personal liability. Miles Jennings, general counsel of the crypto arm of the investment firm Andreessen Horowitz, called the ruling “a huge blow to decentralised governance”, and warned that “any DAO participation, even posting in a forum, could be sufficient to hold members liable”. After the ruling, several decentralised-finance protocols quietly restructured their governance.
The responses: wrapping the DAO in something the law does recognise
The industry reacted the only way it could: if the problem is the absence of legal personhood, the DAO must be given a “legal wrapper” the law recognises. Several models emerged, and comparing them reveals a geography of regulation.
The first to move was Wyoming, which in 2021 created the DAO LLC figure —a limited-liability company with specific provisions for algorithmic governance— and in March 2024 added the DUNA (Decentralised Unincorporated Nonprofit Association), in force since 1 July 2024. The DUNA recognises the DAO as a legal entity with limited liability, but requires a minimum of 100 members, a nonprofit purpose, a Wyoming registered agent and governance documentation. Its setup cost is around 23,000 dollars plus annual fees.
The second pole is the Marshall Islands, which through the 2022 DAO Act, its 2023 amendment and the 2024 regulations built a specific regime granting DAOs full legal personhood: they can own property, contract, sue and be sued in their own name. Nonprofit DAOs operate tax-free; for-profit ones pay 3% on gross revenues. One striking feature: the regime allows the certificate of formation to reduce or eliminate fiduciary duties among members, so they cannot sue each other over how they voted.
To these are added offshore foundations —in jurisdictions such as Panama or the Cayman Islands— used for years as the classic wrapper for for-profit projects, and vehicles in United Arab Emirates free zones. The resulting picture is one of a market of legal domiciles competing to attract DAOs by offering different combinations of protection, taxation and requirements.
| Model | Legal personhood | Limited liability | Notable restriction / cost |
|---|---|---|---|
| No wrapper | No | No (de facto partners) | Risk of unlimited personal liability |
| DAO LLC / DUNA (Wyoming) | Yes | Yes | DUNA: minimum 100 members, nonprofit; ~23,000 USD setup |
| DAO LLC (Marshall Islands) | Yes | Yes | 3% on gross revenues (for-profit) |
| Offshore foundation | Yes (of the foundation) | Yes | Indirect structure; real control must be documented |
On top of all this is a complementary layer of protection the market itself recommends: directors-and-officers (D&O) insurance, with annual premiums that, depending on the size of the DAO’s treasury, range from 50,000 to 500,000 dollars.
The gap no wrapper fully closes
It is worth setting out the two readings of this moment with comparable weight, because the question is far from settled.
On one, optimistic reading, the maturation is healthy and expected. Its defenders hold that every new organisational technology goes through a phase of friction with the law until the latter creates the right categories, and that the appearance of the DAO LLC, the DUNA and the Marshall Islands regimes shows the legal system is adapting relatively quickly. Under this reading, rulings like Lido’s do not kill the DAO idea, but push it to professionalise: the DAO that adopts a legal wrapper gets the best of both worlds —decentralised governance on-chain, limited liability off it. “The code is law” is reinterpreted as “the code operates within the law”, a more mature and sustainable position.
On the contrary reading, the rulings reveal a contradiction the wrapper only disguises. Its defenders argue that the DAO’s original ideal —a truly ownerless organisation, with no central point of control or capture— is incompatible with the legal requirement that there always be someone identifiable to answer. The moment a DAO adopts a registered agent, a domicile in Wyoming or the Marshall Islands and a documented minimum of members, it ceases to be the radically decentralised thing it promised to be and becomes a company with extra steps. For this position, the legal wrapper does not close the gap between the ideal and the law: it manages it, in exchange for giving up what made the DAO distinct. And an underlying open question remains: if most DAOs end up needing a traditional wrapper to function without exposing their members, how do they differ, legally, from the companies they claimed to surpass?
It is not for this outlet to rule which reading will prevail. It is to note the fact both share: the promise that code could replace law has collided with an old and stubborn truth of the legal system —that behind every organisation acting in the world, sooner or later, someone has to answer. DAOs have not repealed that rule. They have discovered, case by case and ruling by ruling, how much it costs to ignore it.
Methodological note. The cover chart estimates each model’s “relative personal-liability exposure” on a 0-100 scale, where a higher value means more risk to members’ individual assets, not a monetary measurement. The estimate weighs two documented factors: whether the model grants legal personhood and limited liability (the absence of a wrapper, where members can be treated as de facto partners, receives the maximum risk value), and the robustness and recognition of the regime. Models with a recognised wrapper (Wyoming, Marshall Islands, offshore foundation) drastically reduce that exposure, with minor differences among them according to their requirements. The cost data (≈23,000 USD setup in Wyoming, 3% on gross revenues in the Marshall Islands, D&O premiums of 50,000–500,000 USD), the requirements (minimum 100 members in the DUNA) and the dates come from sector legal guides and the coverage of the cited cases. The court rulings (bZx, Ooki, Lido) come from the legal coverage of each case. Data cutoff: 23 May 2026.