The DAO of Joint Liability – Harris Sliwoski LLP
One month into a new administration inaugurated in part through the sale of billions of dollars of Trump meme coins, it should come as no surprise that the new administration is expected to put a full stop on the executive branch’s prodigious output of crypto and NFT enforcement actions in recent years. Even so, the courts continue to wrestle with the challenges of regulating an industry designed to resist regulation.
A decision from the U.S. District Court in Northern California published late last year has set off a bit of a panic in the industry, particularly among those who have tended to take refuge in the perceived lack of effective enforcement mechanisms to reign in DAOs, or decentralized autonomous organizations, which are often the promoters of new coin or token offerings. There are over 10,000 known DAOs, with millions of token holders around the globe with a combined total treasury in the billions. Not only did the court in the Samuels v. Lido DAO case decide to answer the question of whether a DAO could be sued at all in the affirmative, by finding the Lido DAO to be a general partnership under California law, but it also attributed to the partners—the individual token holders of the DAO—joint and several liability for the debts of the DAO itself.
By relying on general partnership law and attaching joint and several liability for partnership debts to individual partners, the federal court for the district that includes Silicon Valley relied on legal principles with origins in the Middle Ages to bring DAOs, which were supposed to represent a new form of decentralized entity governance and are intertwined with huge crypto and NFT projects, to their virtual knees.
The case was filed by an investor who purchased tokens issued by Lido DAO and lost money when he sold the tokens at a loss a few months later. The tokens should have been registered with the SEC because they were securities, argued the plaintiff, and the DAO’s failure to do so created liability under U.S. securities law.
Lido DAO is a decentralized autonomous organization providing staking services for the Ethereum blockchain, where transactions are verified through “proof-of-stake,” as validators confirm new blocks in exchange for Ether rewards. Validators must stake their own Ether to participate, with rewards proportional to their stake, and risk losing their stake if they validate dishonestly. Due to the 32 Ether minimum required to be a validator (over $80,000 as of February 2025), there is a lucrative market for firms offering services that pool users’ cryptocurrency and share the fees proportionately.
There is a decent argument that the Lido DAO tokens are securities (see this article of mine for a explanation why). There is no dispute as to whether the tokens were registered with the SEC; they weren’t. Staking Ether has become commonplace. DAOs have been around since no later than 2016, when (the confusingly named) The DAO raised $150 million of Ether in a public offering. In what became the first major treasury raid of the blockchain era, The DAO collapsed after a security flaw infamously allowed a still officially anonymous attacker to drain almost half The DAO’s treasury.
What makes the Lido DAO case different is the extent to which it directly addresses the potential liability of individual token holders in a DAO for liabilities of the DAO itself and, maybe even more significantly, how it may portend the death of the dream of a truly decentralized entity, made of and governed by code, beyond the reach of real-world regulators. After this decision, and unless and until the court’s legal theories change materially during the course of the litigation, it essentially is malpractice per se for an attorney to allow a DAO client to exist and operate outside the protective “wrapper” of a domestic or foreign LLC or other corporate form. For many in the blockchain community, however, this is kind of like telling a group of jazz musicians that they cannot jam unless they first compose a tune and file it with the copyright office.
In a little black book (literally) called Blockchain Governance, part of The MIT Press’ Essential Knowledge Series, the authors introduce the concepts “rule by code” and “rule of code” to illustrate the difference between the centralized online platforms we know as Web 2.0, generally owned and operated by giant technology companies and susceptible to the influence of governments, special interests and corruption, and the decentralized, public and “permissionless” blockchain network of Web 3.0, where code-based rules, i.e., smart contracts deployed via blockchain, “run themselves.”
“Rule of code” is pretty good shorthand for the main argument made to the court by Lido DAO, i.e., the DAO is nothing but software, a decentralized, publicly available “protocol” used to “deploy” an autonomous smart contract system. It cannot be sued, according to this view, because it does not exist as an “entity run by people.”
The court rejected this argument as solipsistic clap trap, highlighting, among other things, Lido DAO’s 70 employees, its functioning token-based governance system and the $280 million in staking profits sitting in its treasury. The well-heeled investor-defendants, including Andreesen Horowitz, Paradigm Operations and Dragonfly Digital Management, who had invested some $100 million in the aggregate to purchase enough governance tokens to significantly influence, if not control, the outcome of DAO votes, did not, as the court wrote, have the “audacity” to argue the DAO is nothing but software. Instead, they argued, somewhat circularly, “whatever Lido is, it is not a general partnership, which means that the investors can’t be liable as general partners for Lido’s conduct.”
The essentially unlimited joint and several liability of partners in a general partnership for the debts of the partnership is basically the third rail of corporate liability, a deadly hazard to be avoided at all costs. It’s the kind of thing most law students learn about by their second year yet rarely see in practice. Lawyers who create business entities for clients mostly form limited liability companies and corporations, they tend not to intentionally create general partnerships. Indeed, most general partnerships arise by implication, by what lawyers like to call “operation of law,” which is why business litigators tend to know more about general partnerships than their transactionally focused colleagues.
Two or more individuals or entities who go into business together with the intention of sharing profits generated by their collective efforts are partners in a general partnership if they have not made arrangements to exist as another form of entity. If there is no written partnership agreement, statutory rules dictate what amounts to a default partnership agreement. About 80 percent of U.S. states and territories have enacted a version of the Revised Uniform Partnership Act of 1997 (“RUPA”), while the rest have seen fit to stick with a version of its predecessor, the Uniform Partnership Act of 1914.
California has a version of RUPA, which the court applied according to its own precedent from 2 years prior when, in a case the Commodity Futures Trading Commission brought against another DAO, the court followed the rule that the law of the state in which a federal court sits applies to parties who are not individuals or incorporated entities.
The Lido court had no problem finding a general partnership was formed under California law when Lido DAO’s founders united to create and operate an Ether staking service for profit. As the court was evaluating a motion to dismiss, where Samuels need only demonstrate his claims are legally and factually viable to them to survive this initial attack, it unsurprisingly acknowledged that determining the partnership’s exact contours, including the full list of general partners and the precise nature of their liability, required further discovery.
This did nothing to help the cause of the investors, however. The extent of their participation in Lido DAO’s governance, especially by exercising significant voting rights associated with “LDO” tokens, gave them an instant seat at the table. While later rulings in the case could conceivably pare down the list of general partners so as to limit the liability of some token holders, there is no escape for investors who promoted and advised the DAO and owned enough tokens to essentially control the DAO alongside its founders. If the DAO is a general partnership, then they are general partners, with seemingly unlimited joint and several liability for partnership debts. Like securities law violations. And like treasury raids, which can be even worse.
In 2023, Nouns DAO, known for its innovative governance model granting one vote per NFT, experienced a massive treasury drain. An investor known as DCF GOD (in reality, a group of members acting together) exploited an arbitrage opportunity by acquiring a sufficient number of Nouns NFTs at a comparatively low price, which afforded DCF GOD the voting power to propose and successfully enact the liquidation of the DAO’s treasury. DCF GOD had discovered the DAO’s core weakness—the treasury’s value wasn’t aligned with the market cap of the governance tokens, i.e., the value of the funds in the treasury significantly exceeded the value of all tokens in circulation multiplied by their current market value. The raiders were able to pull $27 million out of the Nouns treasury, shaking the DAO to its core and tainting its reputation.
The raid on the Aragon DAO’s treasury, also in 2023, was even worse. Formed by idealistic coders in 2017 as a Swiss nonprofit with a goal of becoming a decentralized nation-state, Aragon grew to have a treasury of some $200 million and became a leader in Ethereum’s grant ecosystem. When some of its planned projects didn’t pan out, however, the value of token tied to the governance of those projects dropped, creating an arbitrage opportunity. Once the raiders obtained 51% of the tokens, they were able to drain $163 million from the DAO’s treasury, over the DAO founders’ objections, after which the DAO summarily collapsed.
It does not require much imagination to see how a treasury raid in a world where DAOs are treated like general partnerships could lead to litigation where individual token holders whose investments flounder after a treasury raid sue the institutional token holders who got in early for their lost investments. While the notion has a sort of moral resonance, it would amount to an existential threat to very idea of decentralized governance and ownership via blockchain.
Miles Jennings, general counsel and “head of decentralization” at a16z Crypto, a venture capital fund launched by Andreessen Horowitz, posted on X immediately following the announcement of the court’s Lido decision. Now, he said, “any DAO participation (even posting in a forum) could be sufficient to hold DAO members liable for the actions of other members under general partnership laws.”
Jennings is correct. And also incorrect.
He’s right that in a true general partnership, every general partner is jointly and severally liable for the debts of the partnership. This model has roots in 13th century Europe, specifically the form of entity known as a compagnia, in the city-states that would eventually become Italy. Initially, the compagnia—or “company,” or “fellowship”— consisted of partners related to each other. Also known as “table companions,” these familial partners inspired trust among the people they did business with, in that each of the folks around the table would stand behind the partnership and essentially guarantee its debts, irrespective of which partner was representing the partnership when it made the deal. Even as the compagnia eventually expanded to include partners not related to each other, membership still required permission and the principal of joint and several liability remained.
Jennings is not quite right because he overstates for effect what the Lido court—and other courts to consider essentially the same issues—seem to be telegraphing about where this moving target of a legal theory might end up once these cases make their way to resolution. The courts seem to grasp how certain aspects of DAO governance, including token-weighted voting, make it problematic to impose joint and several liability on every token holder (let alone on any random person who posts in a DAO forum). It may be that future developments in the case narrow the pool of general partners, potentially reducing liability for some token holders, though the investors who actively promoted and advised the DAO—while holding enough tokens to wield significant control alongside the founders—undoubtedly will remain liable.
The investor defendants tried to convince the court that the Lido DAO could not be a general partnership if anyone who bought a token could be a partner, and if any partner could sell its tokens to someone else on the secondary market—particularly if, at the same time, not every partner necessarily has the same degree of liability for the debts of the partnership. Essentially, they were arguing, the court was invading the sacred purview of the fellowship. Or, in other words, if we can’t pick our own table companions, how can we be liable for what some of them do?
Just as the very existence of a partnership itself can be inferred from facts and circumstances, so can the existence—and terms—of a partnership agreement. And a partnership agreement, whether written, oral or—even squishier—one that arises by implication, can vary what are referred to as the “default” partnership rules, including the rules about the relative liability of the partners and how one joins or leaves the partnership.
The court did exactly this, inferring that Lido DAO’s founders likely agreed, explicitly or implicitly, to modify California’s default partnership rules regarding partner admission and exit. The DAO’s structure, with its token-based governance, suggested an intent to create a partnership with flexible membership based on token ownership.
So, what’s a DAO to do, then?
As Jennings posted immediately following the Lido decision, “It’s time to DUNA.”
A DUNA is a Decentralized Unincorporated Nonprofit Association. It’s one of two forms of legal wrapper available to DAOs under legislation enacted by Wyoming. DUNAs, introduced in 2024, must be designed to advance nonprofit objectives and have at least 100 members. While they cannot distribute profits or dividends, DUNAs may “reasonably” compensate management, members, and third parties for services, and they may enter into contracts, including buying and selling property. A DUNA’s members, managers, or administrators cannot be held liable for a breach of a DUNA’s contract because of their association with the DUNA.
The Wyoming DAO LLC statute, enacted in 2021, was the first of its kind in the U.S. It grants DAOs legal status as a distinct type of LLC quite different from traditional LLCs. DAO LLCs can be largely, if not entirely, “algorithmically managed,” meaning, for example, the Operating Agreement can be a smart contract. A smart contract managing the LLC must be capable of being updated by the members, pursuant to a 2022 amendment to the DAO statute. Interestingly, though, a DAO LLC may not be manager managed. One reason is to avoid invocation of the “efforts of others” prong of the infamous Howie test used to identify the sub-species of securities known as investment contracts.
DAO LLCs also have the flexibility to restrict or eliminate fiduciary duties, transfers of membership interests, member withdrawal, return of member capital contributions, and dissolution. Among the concepts floating around the DAO community are novel voting mechanisms like “quadratic” or “reputational” voting, as well as building into a DAO’s structure mechanisms for what are known as “rage quits,” custom provisions in the DAO agreement allowing a member to exit the DAO with some or all of its money, as a disincentive for unhappy token holders to foment and carry out treasury raids.
Larger, more complex DAOs may benefit from a layered approach to legal wrappers, using multiple legal entities for different functions or activities. A base-layer wrapper can encompass the larger DAO community, for example, providing a foundational legal entity, while operating-layer wrappers can be designated for specific activities or transactions.
Finally, for now, opportunities for jurisdictional and regulatory shopping are on the rise. Not only are other states, including Utah, Vermont and Tennessee, currently in on the DAO statute action, but also a number of foreign jurisdictions, like Switzerland, the Cayman Islands, and the Marshall Islands, offer legal structures suitable for DAOs.