← Back to Archive
Programmable Custody Infrastructure
Issue #01618 min read

Programmable Custody: The Invisible Infrastructure That Decides Who Really Owns What When Money Becomes Code

On October 29, 2025, Paxos accidentally created $300 trillion in PYUSD tokens - then burned them twenty minutes later. You held your keys the entire time. But did you own your crypto? When smart contracts mediate ownership, custody becomes programmable. And whoever writes the code holds the real power.

Last updated:

TL;DR

  • Paxos accidentally minted $300T PYUSD (3x global GDP) on October 29, 2025, then burned it via admin keys - proving smart contract issuers can override user private keys. 60% of institutions use third-party custodians despite self-custody capability because control fragments across protocol developers, governance token holders, and multi-sig signers
  • Programmable custody through DeFi protocols, multi-sig wallets, upgradeable contracts, and cross-chain bridges transforms ownership into conditional permission - you don't own assets, you own claims mediated by code that can pause, upgrade, or restrict access without your approval
  • GENIUS Act (July 2025), UK FCA CP25/14, EU MiCAR, and Basel Committee standards treat smart contract deployers as regulated custodians if they retain admin keys - forcing choice between compliance licensing or removing emergency controls that protect against exploits
  • Legal gap: Courts haven't determined if smart contract bugs create liability for developers, whether DAOs are suable entities, or who owes assets when cross-chain bridges fail. Regulatory response assigns custody responsibility to identifiable parties who can be held accountable when code fails

No agenda. No noise. Just clarity.

Get the MCMS brief - digital assets, AI, and law explained with evidence, not hype.

Join 1,000+ professionals. Unsubscribe anytime.

Reader Navigation Guide

Jump to sections relevant to your role

Reader RoleRelevant Sections
Legal & Compliance
Click to view sections →
Custody Licensing — GENIUS Act, UK FCA CP25/14, MiCAR requirements
Liability Assignment — Protocol developer responsibility for losses
Shift from Possession to Permission — How smart contracts redefine legal ownership
Risk Management & Security
Click to view sections →
The Four Layers of Programmable Custody — Escrow, multi-sig, upgradeable contracts, bridges
Circuit Breakers — Mandatory admin keys and emergency pause functions
The Custody Stack — Protocol developers, wallet providers, treasury yield game
What Happens if We Get This Wrong — Unanswered legal and regulatory questions
Developers & Infrastructure
Click to view sections →
Upgradeable Contracts — What happens when ownership rules can be changed
Smart Contract Escrow — Who controls assets locked in DeFi protocols
Cross-Chain Bridges — Custody fragmentation across blockchains
Circuit Breakers — ERC-7265 and mandatory pause functions
Institutional Custody
Click to view sections →
Why "Not Your Keys, Not Your Coins" Doesn't Apply — Paxos $300T incident, custody reality check
Multi-Signature Wallets — When 2-of-3 signatures are required, who holds power?
The Custody Stack — Why 60% of institutions use third-party custodians
Custody Licensing — Regulatory requirements for institutional grade custody
Treasury & Finance
Click to view sections →
How "Owning Crypto" Became "Having Permission" — From ownership to conditional permission
The Treasury Yield Game — DeFi deployment custody considerations
Legal vs Cryptographic Control — What do token holders actually own?
How Decentralized Is Custody Really? — Code upgrades, pauses, migrations
Policy & Regulatory Analysis
Click to view sections →
Three Regulatory Vectors — Custody licensing, liability, circuit breakers
DeFi Promises Disintermediation — Smart contracts as new intermediaries
Treating Smart Contract Deployers as Custodians — GENIUS Act, UK FCA, MiCAR frameworks
What Happens if We Get This Wrong — Unresolved legal questions, enforcement gaps

This navigation framework is exclusive to MCMS members. Share this article to provide colleagues with the same analytical depth.

On October 29, 2025, Paxos - the regulated issuer behind PayPal's U.S. dollar stablecoin - accidentally created $300 trillion worth of PYUSD tokens in a single blockchain transaction. Three times global GDP materialized on-chain in seconds.

Twenty minutes later, those tokens were gone. Burned through the same administrative controls that created them.

If you held PYUSD in your self-custody wallet during those twenty minutes, you controlled your private keys. But did you own your crypto?

I'm sure that when this came out, more than one compliance officer asked "who has the authority to fix this?" after deciding this is too big to be a hack.

The Paxos incident reveals what compliance officers, treasury managers, and wealth advisors need to understand about digital asset custody in 2025: "your keys, your crypto" is a marketing slogan, not a legal framework. The crypto industry's foundational mantra - "Not Your Keys, Not Your Coins" - doesn't actually mean your keys guarantee your coins either.

When smart contracts mediate ownership, custody becomes programmable. And whoever writes or controls the code holds the real power.

Diagnosis - Why "Not Your Keys, Not Your Coins" Doesn't Apply When Code Holds the Keys

The crypto industry built its identity on a simple promise: self-custody eliminates intermediaries. Hold your private keys, control your assets. No banks, no brokers, no trusted third parties. The blockchain verifies ownership cryptographically, and mathematics replaces institutional trust.

This narrative worked when Bitcoin meant holding keys to addresses on a transparent ledger. But institutional crypto in 2025 looks nothing like this. Assets sit in DeFi protocols earning yield. Treasury strategies involve multi-signature wallets with third-party co-signers. Tokenized securities live in smart contracts with compliance modules. Cross-chain bridges fragment custody across multiple blockchains.

Sixty percent of institutions now use third-party custodians despite having technical capability for self-custody. Ninety-two percent of large funds rely on institutional custodians.

The reason isn't technological incompetence - it's structural reality. When smart contracts define the rules of ownership, possession of private keys becomes just one variable in a complex custody equation.

The Paxos incident made this explicit. PYUSD holders maintained cryptographic control of their wallets throughout the minting error. Their private keys never changed. Yet Paxos executed a burn function that could have eliminated tokens from any wallet - no signature required, no permission requested. The smart contract's admin keys superseded every user's private keys.

This isn't a bug. It's how programmable custody works.

Reframe - The Shift from Possession to Permission: How Smart Contracts Redefine Ownership

Traditional finance separates custody from control through legal frameworks. When you deposit cash in a bank, the bank holds the money (custody) but you retain the right to withdraw it (control). Law defines the relationship. Regulators enforce it. Courts resolve disputes.

Cryptocurrency promised to collapse this distinction. If you hold the private keys, you have both custody and control - mathematically guaranteed, no legal system required. The phrase "not your keys, not your coins" captured this elegantly: cryptographic possession equals ownership.

Smart contracts broke that promise.

When assets move into smart contracts - whether for DeFi lending, DAO treasuries, or tokenized securities - ownership becomes conditional. You don't own the asset; you own a token that represents a claim on an asset held by code. That code defines the rules: when you can withdraw, under what conditions, and who can override those rules.

This is programmable custody: ownership mediated by software that can be paused, upgraded, or overridden by parties who control the contract's administrative functions. Your private keys let you interact with the system, but they don't guarantee you control the asset.

The distinction matters legally. In traditional custody, if a bank refuses to return your deposit, you sue the bank. If a smart contract locks your tokens because the protocol developer programmed an exit restriction, who do you sue?

The code? The developer? The DAO that governs the protocol?

The answer remains unclear, which is precisely why this model creates risk for institutional portfolios.

Evidence - The Four Layers of Programmable Custody

Smart Contract Escrow: Who Controls Assets Locked in DeFi Protocols?

When you deposit USDC into a lending protocol like Aave or Compound, your tokens don't stay in your wallet. They transfer to a smart contract that holds them in escrow. In exchange, you receive a derivative token (aUSDC, cUSDC) representing your claim on the underlying asset plus accrued interest.

You now depend on the smart contract's code to withdraw. If the protocol has an emergency pause function - and most institutional-grade protocols do - someone controls the key that can freeze all withdrawals. That someone is usually the protocol development team or a multi-signature wallet controlled by core contributors.

The Paxos PYUSD incident demonstrated this explicitly. Paxos operates as a regulated trust company with NYDFS oversight, yet the stablecoin's smart contract included administrative functions that allowed the company to mint and burn tokens without user approval.

When the erroneous $300 trillion mint occurred, Paxos didn't need to ask permission from PYUSD holders to reverse it. The code gave them unilateral authority.

This isn't unique to Paxos. Every major stablecoin - USDC, USDT, BUSD (before its shutdown) - includes blacklist functions that let issuers freeze tokens in specific wallets. Regulators require this for sanctions compliance.

But it means "your keys, your crypto" comes with an asterisk: your keys work until the issuer decides they don't.

DeFi protocols face the same structural reality. Duke University research identifies smart contract escrow as a "critical DeFi primitive" but notes that tokens become "permanently custodied" if the contract lacks proper withdrawal mechanisms or if admin keys can restrict access.

When Curve Finance suffered a $62 million exploit in July 2023 due to a Vyper compiler vulnerability, the question wasn't just "how did the hack happen" but "who actually had custody of the funds locked in those pools?"

The liquidity providers? The protocol developers? The governance token holders who could vote on recovery measures?

The answer: custody had fragmented across multiple parties, none of whom could unilaterally access the funds, but all of whom had some degree of control over what happened to them.

Multi-Signature Wallets: When 2-of-3 or 3-of-5 Signatures Are Required, Who Holds the Power?

Institutional treasuries increasingly use multi-signature wallets to prevent single points of failure. A 2-of-3 setup requires two out of three designated keyholders to approve any transaction. This eliminates the risk that one compromised employee or stolen device can drain the treasury.

But it also means no individual has custody. Ownership becomes collective and conditional.

If you're one of three signers on a $50 million corporate Bitcoin treasury, you don't control those funds - you control one-third of the access mechanism. The other two signers must cooperate, or the funds stay locked.

This creates governance risk. What happens if one signer leaves the company? What if two signers collude? What if one signer loses their key?

A professional colleague recently shared this story that gives an interesting POV on this: A mid-sized fund I advised had three signers for their €40M treasury. When their head of operations left abruptly during a restructure, it took six weeks and two law firms to execute a single withdrawal. The Bitcoin was 'theirs' the entire time. They just couldn't access it.

Multi-sig wallets solve the security problem but introduce coordination risk. And in most institutional implementations, at least one of those signers is a third-party custody provider like BitGo, Fireblocks, or Coinbase Custody.

The Bybit hack in February 2025 illustrated the vulnerability. While details remain under investigation, reports suggest the breach involved compromised multi-sig credentials rather than a pure smart contract exploit.

When custody depends on multiple parties correctly securing their keys and following protocols, the "no intermediaries" promise dissolves into committee-based permission structures.

Multi-signature wallets also concentrate power in DAOs and protocol governance. When a DeFi protocol's treasury sits in a 5-of-9 multi-sig controlled by core team members, those nine individuals have effective custody over potentially hundreds of millions in user funds, regardless of what the "decentralized" branding suggests.

The Swerve Finance governance exploit in March 2023 showed this risk: an attacker didn't hack the code - they legally acquired governance tokens and used them to control the multi-sig, draining the treasury through legitimate voting mechanisms.

Who had custody before the attack? The multi-sig holders. Who had custody after? Still the multi-sig holders - just different ones. The code executed exactly as designed. The users who thought their funds were secure in a "trustless" protocol learned that trust had simply shifted from banks to governance token holders.

Upgradeable Contracts: What Happens When the Code That Defines Ownership Can Be Changed?

Blockchain's value proposition rests on immutability: code executes as written, no one can change the rules retroactively, and ownership records can't be altered.

Except when they can.

Most institutional-grade smart contracts use upgradeable proxy patterns. The user interacts with a proxy contract that delegates logic to an implementation contract. If the implementation has a bug or needs new features, the proxy can point to a new version - effectively rewriting the rules while maintaining the same address and user balances.

OpenZeppelin, the industry standard for smart contract development libraries, explicitly documents proxy patterns as best practice for production systems. The rationale is sound: code will have bugs, regulations will change, and protocols need the ability to adapt. But upgradeability fundamentally contradicts immutability.

When a smart contract is upgradeable, whoever controls the upgrade mechanism has custody in the truest sense: they can change the rules of ownership. That control typically sits with a multi-sig wallet held by the protocol team, a DAO governance process, or a designated admin address.

This means your tokens in an upgradeable contract exist under conditions that can change. The yield rate can be modified. Withdrawal restrictions can be added. Token balances can be migrated to a new contract.

All without your explicit approval - your only recourse is to exit before the upgrade occurs, assuming you notice it's happening.

The risk isn't theoretical. Multiple DeFi protocols have used upgrade mechanisms to change tokenomics, alter governance structures, or implement emergency security measures. In each case, the code was "immutable" until the moment it wasn't. Users had "custody" until the protocol developers decided different rules should apply.

Regulators are now scrutinizing this model. The French banking authority explicitly notes that as finance decentralizes technologically, "re-concentration occurs in less regulated areas" - meaning protocol developers become the new intermediaries, just without the licensing requirements or liability frameworks that govern traditional custodians.

Cross-Chain Bridges: Where Does Ownership Sit When Assets Move Between Blockchains?

When you bridge USDC from Ethereum to Polygon, you don't actually move the tokens. You lock them in a smart contract on Ethereum and receive newly minted wrapped tokens on Polygon.

Custody has split across two blockchains, two smart contracts, and often two separate validator sets that verify the bridge transactions.

If the Ethereum contract has a bug, your original USDC is at risk. If the Polygon contract has a bug, your wrapped tokens could become worthless. If the bridge's signature verification fails - as happened in the Wormhole bridge hack in February 2022, resulting in $320 million stolen - both chains' assets become compromised simultaneously.

Cross-chain custody creates fragmentation without clear legal ownership. If you hold wrapped Bitcoin on Ethereum, who has custody?

The bridge protocol that locked the real Bitcoin? The smart contract that issued the wrapped version? You, the holder of the wrapped token? The validators who process bridge transactions?

The answer is all of them, partially. Custody has been distributed across a technical stack where each component has some control, but no single party has complete authority.

This is programmable custody at its most complex: ownership mediated by multiple layers of code, operated by different entities, subject to separate governance mechanisms, spanning multiple legal jurisdictions.

For institutions, this creates an accounting nightmare. When your treasury holds $10 million in bridged assets, what do you list on the balance sheet? What disclosures do you make about custody risk?

If the bridge fails and assets become unrecoverable, who's liable - the bridge developers, the validators, the original blockchain's protocol team, or your own risk management for choosing that particular bridge? These questions lack clear answers because the legal framework hasn't caught up to the technology. And that gap is where regulatory intervention will arrive.

Follow the Money - The Custody Stack: Protocol Developers, Wallet Providers, and the Treasury Yield Game

Follow the control points, and you'll find the real custodians.

Protocol developers deploy smart contracts and often retain admin keys for upgrades and emergency functions. Wallet providers like MetaMask and Ledger manage the interface between users and protocols, sometimes with their own custody of seed phrases through recovery services.

Governance token holders vote on treasury management, protocol parameters, and upgrade proposals. Validators and node operators process transactions and can theoretically censor specific addresses.

Each layer represents partial custody. No single party can unilaterally move funds, but multiple parties can restrict, freeze, or redirect them through their respective control points.

The treasury yield game makes this explicit. Institutions don't hold idle crypto - they deploy it into DeFi protocols to earn returns. That deployment means transferring custody to smart contracts controlled by protocol teams they've never met, governed by token holder votes they don't participate in, secured by multi-sig wallets they can't audit, and subject to upgrades they won't learn about until after implementation.

Michael Egorov's position in Curve Finance demonstrated the governance concentration risk. As founder, he held enough CRV tokens to significantly influence protocol decisions while simultaneously borrowing $85 million against those tokens from various DeFi platforms.

When Curve suffered the Vyper exploit, the interconnected custody relationships created systemic risk: Curve's smart contracts held user funds, Egorov's governance position influenced the protocol's response, and his leveraged positions across multiple platforms meant a Curve failure could cascade into liquidations elsewhere.

Who had custody during this period? Legally, users held their LP tokens. Practically, the protocol developers controlled the vulnerable code, Egorov influenced governance decisions, and the lending platforms held his CRV as collateral with liquidation rights.

Custody had become a network of interdependent claims, not a clear property right.

This is why sixty percent of institutions use third-party custodians despite having self-custody capability. They're not paying for key management - they're paying for a single legal entity they can sue when custody becomes disputed.

Traditional custodians like Coinbase Custody and Fidelity Digital Assets don't eliminate programmable custody risk, but they consolidate liability. If assets in a DeFi protocol become frozen, the institution sues its custodian, and the custodian navigates the technical and legal complexity of recovery.

The custody stack has simply added a traditional financial layer on top of the crypto infrastructure, reintroducing the intermediaries that blockchain was supposed to eliminate.

Power Shift - How "Owning Crypto" Became "Having Permission from a Smart Contract"

The conceptual shift from ownership to permission happens gradually, then suddenly.

You start by holding Bitcoin in a self-custody wallet. You own it - possession and control unified through private keys. Then you deposit it into a lending protocol to earn yield. Now you own a claim on Bitcoin, represented by a derivative token, subject to the protocol's withdrawal rules. Then the protocol implements an upgrade that changes those withdrawal rules. Now your claim exists under conditions you didn't agree to when you deposited. Then a governance vote implements an emergency pause due to an exploit on another protocol. Now your permission to withdraw has been suspended by a vote you didn't participate in, made by token holders you don't know.

At each step, the transaction felt voluntary. You chose to deposit. You accepted the protocol's terms. You could have exited earlier.

But structurally, ownership has transformed into conditional permission - permission granted by code, which is controlled by developers, influenced by governance, and subject to change through mechanisms you may not even be aware of.

Property law distinguishes between possession (physical control) and ownership (legal right). Cryptocurrency attempted to merge them through cryptographic control, making possession proof of ownership. Smart contracts separated them again, but without the legal framework that governs traditional finance.

When Ethereum hard-forked in July 2016 to reverse the DAO hack and return $60 million in stolen funds, the network proved that governance trumps code. The community voted to rewrite the blockchain's history, violating the "code is law" principle to achieve what they determined was a just outcome. Ethereum Classic emerged as the non-forked chain, preserving the original history where the hack stood.

Users learned a lesson: even on supposedly immutable blockchains, community governance can override cryptographic ownership. Your keys give you possession until the network decides they don't. Nearly a decade later, that same dynamic persists in programmable custody - just distributed across protocols instead of base-layer blockchains.

Illusion of Need - DeFi Promises Disintermediation, But Smart Contracts Are Just New Intermediaries

The term "decentralized finance" contains an implicit promise: removing intermediaries from financial transactions. No banks approving loans, no brokers executing trades, no clearinghouses settling transactions. Smart contracts replace all of them with transparent, automated code.

But code doesn't write itself. Smart contracts are designed, deployed, and maintained by developers. Those developers become the new intermediaries - just without the regulatory oversight or liability that governs traditional financial intermediaries.

The Congressional Research Service's analysis of DeFi notes that decentralization "relocates rather than eliminates intermediaries." Instead of regulated institutions, users now depend on protocol developers, governance token holders, multi-sig signers, validators, and oracle providers.

Each represents a control point, a party whose decisions affect whether you can access your funds.

This matters for institutional custody because it redefines counterparty risk. When you deposit funds with a traditional custodian, you assess that institution's financial strength, regulatory compliance, and insurance coverage.

When you deposit funds in a DeFi protocol, you must assess the code quality, the developer team's competence, the governance token distribution, the multi-sig signer reliability, and the upgrade mechanism's security - none of which have standardized evaluation frameworks or regulatory oversight.

The French banking authority's research makes this explicit: DeFi relocates concentration risk to "less regulated areas." You haven't eliminated trust; you've shifted it from regulated banks to unregulated protocol developers.

And those developers often have more power than traditional custodians ever did. A bank can't unilaterally rewrite the terms of your deposit account. A protocol developer with admin keys can upgrade the smart contract and change your withdrawal conditions - no approval process required beyond the governance mechanism they designed.

This is why the "disintermediation" narrative fails under scrutiny. Smart contracts are intermediaries - just software ones instead of institutional ones. And software intermediaries can be more opaque, harder to hold accountable, and impossible to sue when they fail.

The Paxos incident created no financial losses. The erroneous tokens were burned before any market impact, and customer funds remained secure throughout.

But it revealed the central custody question institutions must answer: if Paxos can mint and burn tokens at will, what do PYUSD holders actually own?

Legally, they own tokens issued by Paxos, a regulated trust company. But cryptographically, those tokens exist only because the smart contract says they do - and Paxos controls the smart contract.

If Paxos executes a burn function, the tokens disappear from your wallet regardless of your private keys.

Now extend this to a scenario where the burn isn't intentional:

Who's liable?

The protocol developers claim they merely deployed code and have no ongoing responsibility. The DAO argues it's a decentralized entity with no legal personality, so it can't be sued. The governance token holders claim they voted in good faith based on proposals they didn't fully understand.

The users had custody of their private keys, so they accepted the risk of interacting with smart contracts.

Everyone claims no liability. No one can restore the funds.

This gap between cryptographic control and legal ownership creates unacceptable risk for institutional treasuries. When a company holds $50 million in DeFi protocols earning yield, the CFO needs to know who's responsible if those funds become inaccessible.

"The code is the code" is not an acceptable answer for financial reporting or fiduciary duty.

Recent legal developments suggest courts will assign liability regardless of technical arguments. When protocol developers claim they're just publishing code with no ongoing relationship to users, judges are skeptical.

When DAOs argue they have no legal existence, regulators respond by targeting the identifiable individuals who operate them. When users claim their funds were stolen from a smart contract, courts look for a defendant with assets to recover.

The result is a patchwork of liability assignment that varies by jurisdiction and case. Some courts treat protocol developers as service providers with ongoing duty of care. Others apply securities law, treating governance tokens as investment contracts that create issuer liability.

Still others focus on consumer protection statutes, holding wallet providers liable for inadequate disclosure of custody risks.

For institutions, this uncertainty is untenable. Treasury strategies require clear understanding of custody arrangements and liability in case of loss. Programmable custody provides neither.

The technical architecture fragments custody across multiple parties, and the legal framework hasn't determined how to assign responsibility when the system fails.

This is why regulatory intervention is inevitable - and already arriving.

Counter-Attack - Three Regulatory Vectors

1. Custody Licensing: Treating Smart Contract Deployers as Financial Custodians

The GENIUS Act, signed into law on July 18, 2025, establishes the first comprehensive federal framework for stablecoin custody in the United States. It requires issuers to obtain either a federal license or state trust charter, maintain one-to-one reserves in approved assets, and submit to regular audits.

Critically, it treats stablecoin issuers as custodians - not technology companies publishing code, but financial institutions holding customer assets.

The UK's Financial Conduct Authority published consultation paper CP25/14 in May 2025, proposing a licensing regime for cryptoasset custody that would apply to any entity providing "safeguarding and administration of cryptoassets belonging to another person."

The definition explicitly includes smart contract deployers who retain admin keys or upgrade capabilities. If the consultation proceeds as drafted, anyone who deploys a custody-related smart contract in the UK would need FCA authorization by 2026.

The EU's Markets in Crypto-Assets Regulation (MiCAR) became fully operational on December 31, 2024, requiring crypto asset service providers to segregate client assets, maintain minimum capital, and implement robust custody procedures.

While MiCAR primarily targets exchanges and wallet providers, its functional approach means that protocol developers who control user funds through admin keys or governance mechanisms could be classified as service providers subject to authorization requirements.

These regulations share a common principle: if you control user assets through code, you're a custodian under financial regulation - regardless of whether you call yourself a protocol developer, a DAO, or a decentralized application.

The technical mechanism (smart contracts vs. database entries) is irrelevant; the functional reality (you can restrict, freeze, or redirect user funds) is what triggers licensing requirements.

For protocol developers, this creates an impossible choice:

  • Retain admin keys and emergency pause functions to protect against exploits, and you become a regulated custodian
  • Remove all admin controls to avoid regulation, and you lose the ability to respond to bugs or hacks

Most institutional-grade protocols will choose compliance - which means programmable custody increasingly looks like traditional custody with blockchain characteristics, not a fundamentally different model.

2. Liability Assignment: Making Protocol Developers Responsible for Losses

The Howey test determines whether an asset is a security based on investment of money in a common enterprise with expectation of profit from others' efforts. The SEC has argued that many DeFi tokens meet this definition, making the protocols that issue them subject to securities regulation - including liability for material misstatements and fraud.

If a protocol markets itself as "trustless" while the development team retains admin keys that can pause withdrawals, is that a misleading statement that creates liability?

If governance documentation suggests decentralized control but a small group holds majority voting power, does that constitute fraud?

The SEC's enforcement actions suggest yes.

Beyond securities law, traditional tort principles are being applied to smart contract developers. If a developer deploys code with a known vulnerability that results in user losses, can they be sued for negligence?

Courts in multiple jurisdictions have allowed such claims to proceed, rejecting arguments that deploying open-source code insulates developers from liability.

The Steptoe analysis of smart contract liability recommends that developers implement formal testing procedures, obtain security audits from reputable firms, establish bug bounty programs, and maintain clear documentation of known risks.

These practices mirror the duty of care expected from traditional financial service providers - suggesting that even without explicit regulation, common law is evolving to treat protocol developers as having ongoing responsibility to users.

For DAOs, the liability question becomes more complex. If a DAO votes to implement an upgrade that contains a bug, are the token holders liable? Is the core development team liable for proposing it? What about the security auditor who reviewed the code?

Recent cases suggest courts will pierce the "decentralized" structure and assign liability to identifiable individuals who had meaningful control over the decision.

This creates significant risk for anyone participating in DeFi governance:

  • Voting on protocol upgrades could create personal liability if those upgrades cause losses
  • Serving as a multi-sig signer could make you a fiduciary with duty of care to users
  • Contributing to a protocol as a developer might establish ongoing responsibility for its security

The legal framework is still forming, but the direction is clear: claiming you're just code or just governance won't shield you from liability when users lose funds.

3. Circuit Breakers: Mandatory Admin Keys and Emergency Pause Functions

The crypto industry initially celebrated immutability. "Code is law" meant no one could freeze your funds, no government could seize them, and no institution could restrict your access.

That principle is being systematically dismantled by security requirements and regulatory mandates.

ERC-7265, a proposed standard for DeFi circuit breakers, would require protocols to implement emergency pause functions that can halt contract execution in case of exploits. Real-time monitoring systems would detect anomalous behavior - large unexpected withdrawals, unusual trading patterns, or smart contract calls that match known attack vectors - and automatically trigger the pause mechanism.

For users, this means "your crypto" can be frozen by automated systems responding to threat detection algorithms.

For institutions, it means custody in DeFi protocols is explicitly conditional: your funds are accessible unless the protocol determines otherwise.

Regulatory frameworks are moving toward mandatory circuit breakers. The Financial Stability Board's recommendations on DeFi regulation include requirements for "kill switches" and emergency intervention capabilities.

The rationale is financial stability - preventing a single exploit from cascading through interconnected protocols and creating systemic risk.

But mandatory admin keys fundamentally contradict the decentralization and trustlessness that define crypto's value proposition. If every protocol must have an emergency pause function, every protocol must have someone who controls that function. That someone becomes the real custodian, regardless of what the marketing materials claim.

The Chainlink documentation on circuit breakers explicitly acknowledges this trade-off: security requires centralized intervention capabilities, which means trust isn't eliminated - just relocated to whoever controls the circuit breaker.

For institutions evaluating custody risk, the question becomes: who controls the pause function, how is it governed, and what prevents abuse?

Some protocols are implementing time-locked admin keys with mandatory delay periods, requiring governance votes before any emergency action. Others use multi-sig arrangements with external security firms as key holders. A few are experimenting with algorithmic circuit breakers that trigger based on predefined conditions without human intervention.

None of these solutions eliminate the core problem: programmable custody requires trust in the people who programmed it. And once you acknowledge that trust requirement, you're back to evaluating counterparty risk - just with worse information and less recourse than traditional custody provides.

Speculation - When Code Can Be Upgraded, Paused, or Migrated - How Decentralized Is Custody Really?

The ideological promise of cryptocurrency is custody without trust. The practical reality in 2025 is custody with trust redistributed across protocol developers, governance participants, and infrastructure providers - none of whom are regulated as custodians, most of whom disclaim liability, and few of whom can be effectively sued when things go wrong.

This is programmable custody: ownership mediated by software that can be modified by people you've never met, using processes you don't control, governed by tokens you may not hold, subject to decisions you won't learn about until after they're implemented.

You have your keys, but custody is everywhere and nowhere - fragmented across a technical stack that no single party fully controls and no regulator fully oversees.

For professional asset managers, this creates untenable risk:

The Paxos incident demonstrated what institutional treasurers already suspected: self-custody in a smart contract world is an illusion. You can hold your private keys, control your wallet, and store your seed phrase in a secure vault - but if the protocol can mint, burn, pause, or upgrade your tokens, custody sits with whoever controls those functions, not with whoever holds the keys.

This isn't an argument against blockchain technology or digital assets. It's an argument for intellectual honesty about custody models.

Programmable custody through smart contracts is not the same as self-custody through private keys. It's a different risk profile, with different control points, requiring different evaluation frameworks.

"The real risk isn't that code will fail - it's that when code works exactly as designed, you'll realize you never understood who programmed the rules."

Implication - What Happens if We Get This Wrong

We don't yet know whether courts will treat smart contract interactions as contractual relationships that create enforceable obligations, or as purely technical transactions where users assume all risk.

We don't know if protocol developers who deploy and maintain code will be classified as fiduciaries, service providers, or something entirely new under law.

We don't know whether DAOs will be recognized as legal entities with liability, or treated as unincorporated associations where participants bear individual responsibility.

We don't know if cross-chain custody will be governed by the jurisdiction where the original asset is locked, where the wrapped token is issued, or where the bridge validators are located.

We don't know whether emergency pause functions will become mandatory across all DeFi protocols, or if some will be allowed to remain truly immutable.

We don't know if institutions will ultimately trust programmable custody enough to deploy significant capital, or if they'll demand traditional custodian intermediaries regardless of the underlying technology.

These uncertainties matter because they define risk that can't yet be priced or managed through traditional frameworks. Institutional treasurers, compliance officers, and wealth advisors operating in this environment are making custody decisions with incomplete information about legal ownership, regulatory requirements, and liability exposure.

What we do know is this: "your keys, your crypto" is no longer an adequate framework for understanding custody in a world where smart contracts mediate ownership.

The code that holds assets is written by someone, controlled by someone, and can be changed by someone. Those someones are the real custodians - whether they acknowledge it or not, whether regulators have classified them yet or not, and whether institutions recognize the risk or not.

The regulatory response currently taking shape will force this honesty. Licensing requirements, liability assignments, and mandatory circuit breakers all acknowledge that programmable custody involves intermediaries - just unconventional ones.

The next phase of crypto regulation won't eliminate smart contract custody, but it will assign responsibility for it to identifiable legal entities that can be held accountable when the code fails.

Call-Forward - What We'll Explore Next

The next MCMS issue will examine "The GENIUS Act Stablecoin Reshuffle" - three months after the law became effective, which issuers made the cut, which got shut out, and what this means for the $300 billion stablecoin market.

Because custody isn't just a technical problem. It's a licensing problem, a compliance problem, and a "who gets to issue dollar-backed tokens" problem. And the answer to that last question will reshape who controls programmable money in the regulated era.

If you read this far, you're already ahead of most professionals.

Join 1,000+ readers who get institutional-grade insights - clear, concise, and verifiable.

No spam. Unsubscribe anytime.

If you found this useful, please share it.

Questions or feedback? Contact us

MCMS Brief • Classification: Public • Sector: Digital Assets • Region: Global

References

  1. 1. Latham & Watkins - The GENIUS Act of 2025: Stablecoin Legislation Adopted in the US (July 1, 2025) [Link]
  2. 2. Paul Hastings - The GENIUS Act: A Comprehensive Guide to US Stablecoin Regulation (January 1, 2025) [Link]
  3. 3. UK Financial Conduct Authority - CP25-14: Regulating the Custody of Cryptoassets (May 28, 2025) [Link]
  4. 4. Central Bank of Ireland - Markets in Crypto-Assets Regulation (MiCAR) (January 1, 2024) [Link]
  5. 5. Basel Committee on Banking Supervision - Prudential Treatment of Cryptoasset Exposures (July 1, 2024) [Link]
  6. 6. UNIDROIT - UNIDROIT Principles on Digital Assets and Private Law (May 1, 2023) [Link]
  7. 7. OpenZeppelin - Upgrading Smart Contracts (January 1, 2025) [Link]
  8. 8. OpenZeppelin - Proxy Upgrade Pattern (January 1, 2025) [Link]
  9. 9. Aave - Aave Protocol Governance Documentation (January 1, 2025) [Link]
  10. 10. Financial Stability Board - The Financial Stability Risks of Decentralised Finance (February 1, 2023) [Link]
  11. 11. Bank for International Settlements - DeFi Beyond the Hype (January 1, 2022) [Link]
  12. 12. ISDA - Legal Aspects of Smart Contract Applications (January 1, 2020) [Link]
  13. 13. MIT - Decentralized Finance: On Blockchain- and Smart Contract-Based Financial Markets (January 1, 2021) [Link]
  14. 14. UK Law Commission - Property (Digital Assets etc) Bill (January 1, 2025) [Link]
  15. 15. Ethereum Improvement Proposals - ERC-7265: Circuit Breaker Token Standard (January 1, 2023) [Link]
  16. 16. World Bank - Regulatory Implications of Integrating Digital Assets (January 1, 2020) [Link]
  17. 17. Skadden - Bank Capital Standards for Cryptoassets (August 1, 2024) [Link]
  18. 18. OKX - PYUSD on Stellar: Global Payments Use Case (January 1, 2025) [Link]
  19. 19. Global Legal Insights - Blockchain and Cryptocurrency Laws and Regulations USA (January 1, 2025) [Link]
  20. 20. ISSA - Custody in a Digital World (October 1, 2023) [Link]

SOURCE FILES

Source Files expand the factual layer beneath each MCMS Brief — the verified data, primary reports, and legal records that make the story real.

Regulatory Frameworks for Digital Asset Custody

The GENIUS Act, signed into law on July 18, 2025, establishes the first comprehensive federal framework for stablecoin custody in the United States, requiring issuers to obtain either a federal license or state trust charter, maintain one-to-one reserves, and submit to regular audits. The law treats stablecoin issuers as custodians - not technology companies publishing code - creating legal responsibility for entities that control user funds through smart contracts. The UK's Financial Conduct Authority published consultation paper CP25/14 in May 2025, proposing a licensing regime for cryptoasset custody that would apply to any entity providing 'safeguarding and administration of cryptoassets belonging to another person.' The definition explicitly includes smart contract deployers who retain admin keys or upgrade capabilities, meaning anyone who deploys a custody-related smart contract in the UK would need FCA authorization by 2026. The EU's Markets in Crypto-Assets Regulation (MiCAR) became fully operational on December 31, 2024, requiring crypto asset service providers to segregate client assets, maintain minimum capital, and implement robust custody procedures. MiCAR's functional approach means protocol developers who control user funds through admin keys or governance mechanisms could be classified as service providers subject to authorization requirements. These regulations share a common principle: if you control user assets through code, you're a custodian under financial regulation - regardless of technical implementation.

Smart Contract Custody Mechanisms and Upgradeable Patterns

OpenZeppelin, the industry standard for smart contract development libraries, explicitly documents proxy patterns as best practice for production systems. Most institutional-grade smart contracts use upgradeable proxy patterns where users interact with a proxy contract that delegates logic to an implementation contract. If the implementation has a bug or needs new features, the proxy can point to a new version - effectively rewriting the rules while maintaining the same address and user balances. When a smart contract is upgradeable, whoever controls the upgrade mechanism has custody in the truest sense: they can change the rules of ownership. That control typically sits with a multi-sig wallet held by the protocol team, a DAO governance process, or a designated admin address. This means tokens in an upgradeable contract exist under conditions that can change without explicit user approval. The Paxos PYUSD incident on October 29, 2025, demonstrated this explicitly. Paxos operates as a regulated trust company with NYDFS oversight, yet the stablecoin's smart contract included administrative functions that allowed the company to mint and burn tokens without user approval. When the erroneous $300 trillion mint occurred, Paxos didn't need permission from PYUSD holders to reverse it - the code gave them unilateral authority. This isn't unique to Paxos; every major stablecoin includes blacklist functions that let issuers freeze tokens in specific wallets for sanctions compliance.

DeFi Systemic Risks and International Standards

The Financial Stability Board's February 2023 report on DeFi identifies concentration risk as a critical vulnerability: despite marketing as 'decentralized,' most protocols have identifiable control points where developers, governance token holders, or multi-sig signers can restrict user access to funds. The FSB notes that decentralization 'relocates rather than eliminates intermediaries,' creating new forms of counterparty risk without the regulatory oversight or liability frameworks that govern traditional custodians. The Bank for International Settlements' analysis concludes that DeFi protocols with admin keys or upgrade capabilities function as financial intermediaries regardless of their technical architecture. The BIS explicitly states that custody in smart contracts requires trust in 'the developers who wrote that cryptography,' contradicting the industry narrative of trustless finance. The Basel Committee's July 2024 standards for prudential treatment of cryptoasset exposures require banks to apply 1250% risk weighting to unbacked crypto assets, effectively treating them as highly risky. For institutions, this regulatory stance reflects official acknowledgment that programmable custody involves intermediaries with uncertain liability - creating untenable risk for professional asset managers who must understand custody arrangements and liability in case of loss.

Legal Frameworks for Digital Property and Smart Contract Liability

UNIDROIT's Principles on Digital Assets and Private Law, adopted in May 2023 by 27 member states plus the EU, establish foundational legal concepts for digital asset ownership. The principles recognize that control of cryptographic keys constitutes a form of possession, but they explicitly acknowledge that smart contracts can fragment custody across multiple parties with different degrees of control - requiring new legal frameworks to assign liability when code fails. The UK Law Commission's Property (Digital Assets etc) Bill, published in 2025, proposes treating digital assets as a distinct third category of property alongside tangible and intangible property. The bill addresses the gap between cryptographic control and legal ownership, recognizing that possession of private keys doesn't necessarily equal full ownership rights when smart contracts mediate access. MIT research on blockchain-based financial markets identifies the fundamental tension: cryptocurrency promised to collapse the distinction between custody and control through cryptographic possession, but smart contracts re-introduced intermediation through code. Academic analysis concludes that 'your keys, your crypto' works only for base-layer blockchain assets - once assets move into smart contracts for DeFi, custody becomes conditional on protocol rules that can be modified by developers, creating legal uncertainty about who bears responsibility when systems fail.

KEY SOURCE INDEX

  • GENIUS Act (via Latham & Watkins)US federal stablecoin legislation (July 18, 2025) requiring custody licensing, 1:1 reserves, and treating issuers as regulated custodians - establishing legal framework for smart contract custody
  • UK FCA CP25/14May 2025 consultation proposing custody licensing for any entity providing 'safeguarding and administration of cryptoassets' - explicitly includes smart contract deployers with admin keys
  • EU MiCARMarkets in Crypto-Assets Regulation operational December 31, 2024 - requires asset segregation, minimum capital, custody procedures; functional approach classifies protocol developers as service providers
  • Basel Committee StandardsJuly 2024 prudential treatment requiring 1250% risk weighting for unbacked crypto - reflects regulatory view that programmable custody creates high counterparty risk
  • UNIDROIT Digital Assets PrinciplesInternational legal framework (27 member states + EU, May 2023) recognizing that control of keys = possession, but smart contracts fragment custody across parties with different control degrees
  • OpenZeppelin (Upgrade Patterns)Industry standard smart contract library documenting proxy patterns as best practice - technical foundation for upgradeable contracts where developers can modify custody rules post-deployment
  • Financial Stability BoardFebruary 2023 report identifying DeFi as 'relocating rather than eliminating intermediaries' - decentralization claim masks concentration of control in protocol developers and governance token holders
  • Bank for International SettlementsCentral bank coordination body concluding DeFi with admin keys functions as financial intermediary regardless of technical architecture; custody requires trust in code developers
  • MIT DeFi ResearchAcademic analysis concluding smart contracts re-introduced intermediation through code; 'your keys, your crypto' applies only to base-layer assets, not DeFi protocols with conditional access

Related Reading

Disclaimer: This content is for educational and informational purposes only. It is NOT financial, investment, or legal advice. Cryptocurrency investments carry significant risk. Always consult qualified professionals before making any investment decisions. Make Crypto Make Sense assumes no liability for any financial losses resulting from the use of this information. Full Terms