What Is a Crypto DAO? How Decentralized Organizations Actually Run

Ask ten people what a DAO is and you get three definitions and a shrug. The useful version is narrow. A decentralized autonomous organization is a group that keeps its rules and its money in code, so decisions and payments do not depend on a manager signing off. Membership usually comes from holding a token, votes are recorded in public, and the treasury sits at an address that either a contract or a small group of signers controls.

That structure solves a real problem, namely trust in a middleman, and it creates new ones. What follows is how the machinery actually works, what it has cost when it went wrong, and the questions worth asking before you join one.

What a DAO is, in plain terms

If you have searched for DAO meaning in crypto, most answers say "a business run by code." A more precise version: a DAO is an organization whose governance rules are enforced by smart contracts on a public blockchain, with voting power distributed through a token or a membership share.

Autonomous is the misleading word. Most DAOs are semi-automated at best. Contracts hold funds and execute approved transactions, while people write proposals, argue on forums, review code and do the actual work. A group of twenty contributors with one shared treasury is still a DAO. It is simply not autonomous in the science-fiction sense, and pretending otherwise leads to bad decisions about what can be left to code.

What genuinely separates a DAO from a company is disclosure. The rulebook is readable, the treasury is visible, and every vote is attributable. You can check whether the people who approved a spend actually held the tokens they claimed to.

How a DAO makes a decision

The pipeline is fairly consistent across large protocols. Someone posts a proposal on a governance forum. Discussion and revisions follow. A temperature check or a signal vote happens on Snapshot, an off-chain voting platform launched in August 2020 by Fabien Marino at Balancer Labs. Snapshot is gasless: voters sign a message rather than send a transaction, and the signatures are stored on IPFS, so results are verifiable but not self-executing. A separate on-chain vote then moves the money.

Compound shows the full sequence. Its governance token went live on 15 June 2020, distributing 0.50 COMP per Ethereum block, roughly 2,880 a day, to suppliers and borrowers. Voting weight comes from delegation rather than simply from holding. An address delegated at least 25,000 COMP can open a proposal, voting runs for three days and needs at least 400,000 votes, and anything that passes waits two days in a Timelock before it can be executed. End to end, a change takes at least a week.

That delay is a feature, not bureaucracy. It is the window in which holders can react if a vote turns out to have been bought, or if the code behind a proposal does something the description did not mention.

The treasury is where the power sits

Money is the part that matters. DAOs hold funds in multisig wallets, most often a Safe, or in contracts that release funds only when a specific proposal has passed. The difference between those two arrangements is the difference between a governed organization and a group chat with a bank account.

A multisig needs several keys to move anything, which is good, but the signers are also a real point of centralization. If three of five people can transfer funds without a vote, then the vote is advisory and the signers are the management. Worth checking before you contribute: who the signers are, what the threshold is, whether they are individuals, a foundation or a paid service provider, and whether the wallet is covered by any policy that limits what it can do. Our guide to how multisig crypto wallets work covers the mechanics.

Treasury management is also where most DAOs grow up. A treasury held entirely in the protocol's own token looks large until the token falls, which is why many now convert part of it into stablecoins to fund a runway measured in years rather than price levels.

The DAO that split Ethereum in two

The first large-scale attempt is still the best cautionary tale. The DAO ran a token sale from 30 April to 28 May 2016 and collected roughly 12.7 million ETH from about 11,000 participants, worth somewhere near $150 million at the time and a large share of all ether then in existence.

On 17 June 2016, an attacker used a reentrancy flaw in the contract's split function to move about 3.6 million ETH into a child contract. The design included a delay before funds could be withdrawn, which gave the community time to respond. On 20 July 2016, Ethereum executed an irregular state change at block 1,920,000 that swept roughly 12 million ETH into a withdrawal contract for the original token holders. About 85% of miners followed the new chain. The rest kept mining the old rules, and that chain became Ethereum Classic.

Two conclusions survived. Complexity is attack surface, and a sufficiently large loss will make a community rewrite its own ledger. "Code is law" holds right up to the point where it becomes too expensive to defend.

What the failure changed

The DAO's collapse put governance experiments on ice for two years. When they returned, they came back smaller. MolochDAO was summoned on 14 February 2019 at ETHDenver by Ameen Soleimani, with roughly 22 founding members contributing 100 ETH each when ether traded near $115.

Moloch's contribution was not better voting. It was the exit. Members held non-transferable shares, and the contract added ragequit: any member could burn their shares for a proportional slice of the treasury before a passed proposal executed. Dissent no longer had to win a vote, it only had to leave. That single mechanism, cheap to implement and easy to audit, has been copied into grant programs, investment clubs and working groups ever since.

Governance tokens and the turnout problem

A governance token grants voting weight, and voting weight is the whole game. The optimistic theory is that holders steward the protocol because they own part of it. The observed reality is that most holders do not vote. Five days after Compound's distribution began, its newsletter reported 139 addresses registered to vote, up from 37.

Delegation exists because of that gap. Holders hand their weight to someone who will show up, which produces functional governance and also concentrates power in a handful of delegates and venture funds. Token-weighted voting has a second weakness: if voting power can be borrowed, a proposal can be carried by someone who owns very little but holds a large temporary balance. Flash loans make that cheap to attempt, which is why timelocks and quorum thresholds are treated as necessary rather than optional.

The practical version for a reader is unglamorous. Look at who actually turns up. A DAO where four delegates decide everything is not decentralized, whatever the token distribution chart shows.

When a DAO wants a legal identity

For years, a DAO had no legal existence, which meant that in many jurisdictions its members risked being treated as a general partnership, personally liable for the group's obligations. Wyoming moved first, with a DAO supplement to its LLC law that took effect on 1 July 2021. In March 2024 the state passed the DUNA Act, a structure for decentralized unincorporated nonprofit associations, effective 1 July 2024.

Uniswap is the most visible test. In 2025, UNI holders approved establishing Uniswap governance as DUNI, a Wyoming DUNA. DUNI now files quarterly and annual financial statements with estimates of US federal and state tax liability; the statements for the year ended 31 December 2025 were published in January 2026. Membership follows from holding UNI and participating, by voting, delegating or submitting proposals, rather than from ownership alone.

The trade-off is real in both directions. A wrapper gives the organization a counterparty that can sign contracts, hold assets and limit member liability. It also brings filings, tax exposure and rules that sit outside the chain, which is a strange place for an organization built on the premise that code settles everything.

The Ooki DAO case: voters can be defendants

Legal risk is not theoretical. In 2022 the US Commodity Futures Trading Commission sued Ooki DAO, the successor to the bZeroX protocol, alleging it offered leveraged retail commodity transactions outside a registered exchange, acted as an unregistered futures commission merchant and skipped customer identification and anti-money-laundering obligations.

Ooki DAO did not mount a defence. In June 2023 the court entered a default judgment of $643,542, imposed permanent trading and registration bans and ordered the website shut down. The ruling held that the DAO is a "person" under the Commodity Exchange Act and can be liable for its violations. The founders were separately held liable as controlling persons and as members of the DAO.

The lesson is uncomfortable for anyone who thinks voting is a harmless signal. Token holders who vote on how a protocol operates can be treated as the people operating it.

A DAO is not automatically decentralized

ConstitutionDAO is the cleanest illustration of the model's limits. Formed in November 2021, it raised about $47 million in ether from 17,437 contributors in under a week through the Juicebox crowdfunding protocol, with a single goal: buy one of the thirteen surviving original copies of the US Constitution at Sotheby's.

It lost. The DAO's bid topped out at a $40 million hammer price, Ken Griffin's side went to $41 million, and the final price with buyer's premium was $43.2 million. Organizers said a higher bid would have left too little for insurance, storage and transport. The project wound down on 23 November 2021 and opened refunds at the original rate, minus Ethereum fees. Weeks later roughly $23 million was still unclaimed, partly because gas costs swallowed small balances. The median contribution was $217, and some contributors paid more in fees than they had donated.

Two transferable points. A single-purpose DAO has no second act when its purpose fails, and transaction costs quietly decide who can afford to participate and who can afford to leave.

How to judge a DAO before you join or vote

• Who can submit a proposal, and what token balance or delegation does it take?
• Is there a timelock, how long is it, and can anyone cancel a queued proposal?
• Who holds the treasury keys, how many signatures are required, and are the signers named?
• Does voting happen only on Snapshot, or does it end in an on-chain execution that actually moves funds?
• How concentrated is delegated voting power, and how many addresses voted on the last three proposals?
• Can members exit with their share before a decision executes, or is leaving a sale into a thin market?
• Is there a legal entity, and what does membership expose you to?

Most of these are answerable in twenty minutes of reading. Failing two or three is not automatically disqualifying, but failing them silently is how people end up surprised by a spend they never saw coming.

What DAOs are actually good at

A DAO is a governance layer, not a company replacement. It is good at pooled money with visible rules, at grant programs that need a credible neutrality story, at coordinating contributors across borders without an incorporation in each country, and at making changes that cannot be slipped through quietly.

It is weak at speed, at discretion, at anything that depends on a signed contract with a counterparty who wants someone to be responsible, and at protecting people who hold small balances from costs that are fixed in absolute terms. Reading the treasury and the last dozen votes tells you more about how a DAO behaves than its manifesto does.