Security incidents have been almost constant across the crypto industry in recent months. Further reading: “From Binance to PHP2Shell: Web Security Must Be Redefined for the AI Era.”
From multibillion-dollar losses at trading platforms to attacks on wallet supply chains and third-party service providers, as well as smart contract exploits and onchain theft, these incidents have repeatedly highlighted the same point: crypto-asset security depends on more than the reliability of the underlying blockchain. It also depends on who holds the assets, who controls the keys, and how many layers of trust exist across the service chain.
The same is true for ETH holders who participate in staking.
Ethereum has established a relatively mature native staking mechanism. By staking ETH, users can become validators, help secure the network, and earn protocol rewards. But as exchange-based custody, liquid staking, and professional node services have created more ways to participate, an easily overlooked question has emerged:
Different ways of staking the same ETH can have fundamentally different asset-control structures and security boundaries.
If ETH is handed to a third party, who controls withdrawals? If a professional service provider operates the validator, can the user still independently control the principal? And what additional risks arise when moving from native staking to liquid staking?
Answering these questions may be more important than simply comparing the APYs offered by different staking products.
1. The First Line of Staking Security: Who Controls Withdrawals?
To understand ETH staking security, we first need to understand who controls an Ethereum validator.
Ethereum uses a proof-of-stake (PoS) consensus mechanism. By depositing at least 32 ETH into the deposit contract, a user can activate a validator, participate in block proposals and attestations, and earn rewards under the protocol’s rules.
A validator, however, is not controlled by a single key. Ethereum separates its core permissions into two distinct categories:
- Signing Key: Used for the validator’s day-to-day duties, including signing block proposals and attestations. It represents the operational authority required to participate in network consensus.
- Withdrawal Credentials: Determine the address to which the staked principal and rewards can ultimately be withdrawn. For validators using execution-layer withdrawal credentials, this address is typically a self-custodial wallet controlled by the user and represents ultimate control over the assets.
The distinction is clear: the signing key determines whether the validator can perform its duties, while the withdrawal credentials determine where the funds can ultimately go.
This means that the party operating a validator does not necessarily need to control the underlying assets.
Once this separation is understood, the fundamental difference between custodial and non-custodial staking becomes much clearer.
With fully custodial staking, both forms of control are generally concentrated in the platform. Users deposit ETH with a trading platform or custodian, which operates the validators and controls withdrawals. What users see on the front end is an account balance. Whether they can successfully redeem their assets depends directly on the platform’s fund management, withdrawal rules, and financial condition.
With non-custodial staking, operational authority and withdrawal control are separated. The user retains control of the withdrawal address through their own wallet, while a professional service provider holds the signing key and manages day-to-day validator operations. Even though the provider handles the hardware, networking, and maintenance, its operational permissions do not allow it to transfer the user’s staked principal to itself.
The key value of this model is a smaller trust boundary: users do not have to give up withdrawal control, nor do they have to bear the full technical burden of running a validator themselves.
Non-custodial staking does not, however, eliminate risk.
The signing key still affects validator performance. If a service provider suffers a key compromise, signs conflicting messages, or experiences a serious outage, the validator may still incur protocol penalties.
The first step in assessing any staking solution is therefore to determine whether withdrawal control is strictly separated from operational authority—and whether users genuinely retain control over recovering their assets.
2. How Ethereum’s Native Protocol Protects Staked Assets
Once key control is understood, another important question follows: why does direct participation in Ethereum's native staking mechanism offer a simpler security structure than staking through third-party derivative products?
The main reason is that native ETH staking is built directly on Ethereum’s consensus protocol, without requiring additional intermediary layers.
When users participate in native staking, their ETH enters Ethereum’s validator system directly. Rewards and penalties are administered automatically under the protocol’s rules.
Native staking does not require a third-party pool to aggregate funds or the minting of a derivative token, such as a liquid staking token (LST), to represent the user’s position. This removes several additional layers of risk, including third-party smart contracts, protocol governance, and potential deviations between a derivative token's market price and the value of the underlying assets it represents.
Within the native protocol, Ethereum establishes clear security and exit boundaries through three main mechanisms.
First, the protocol uses rewards and penalties to regulate validator behavior.
Validators must remain online and submit block proposals and attestations as required. If a validator experiences an outage or remains offline for an extended period, it may miss rewards and incur balance penalties.
More serious violations, such as signing conflicting blocks or attestations, may trigger slashing. Part of the validator’s staked ETH may be destroyed, and the validator may be forcibly removed from the network.
The purpose of this mechanism is to ensure that validators assume real responsibilities in exchange for staking rewards. Any attempt to undermine network consensus carries a real economic cost.
Second, withdrawal control can remain independent of validator operations.
This is one of native staking’s most important security features. Even if a validator’s signing key is compromised, an attacker cannot simply use that key to withdraw the staked principal, provided that the user continues to control the withdrawal address securely.
After the Pectra upgrade, the withdrawal address can submit a validator exit request through the execution layer without using the validator signing key. The exit and withdrawal remain subject to Ethereum’s protocol rules and queues.
Finally, native staking exits remain governed by protocol rules.
Native staking supports both reward withdrawals and full exits, but funds do not return as quickly as an ordinary wallet transfer. After an exit is initiated, the validator must enter Ethereum’s network-wide exit queue. The waiting time depends on how many other validators are exiting at the same time. Once the process is complete, the remaining principal and rewards are automatically transferred to the designated withdrawal address. Further reading: “Why Does Native Ethereum Staking Come With a Month-Long Queue?”
From this perspective, native staking is not valuable solely because it generates ETH rewards. It also provides a settlement framework whose rules are public, whose state can be verified onchain, and whose settlement does not depend on the discretion of a centralized institution.
3. Four Main Models: Where Are Their Security Boundaries?
When key control and protocol dependencies are considered together, the four main ETH staking models differ primarily in where they place risk.
1. Fully Custodial Staking: Centralized Exchanges and Custodians
Staking services provided by exchanges and custodians generally have the lowest participation threshold. They support smaller deposits and offer an experience similar to conventional financial products, without requiring users to manage complex technical operations.
The trade-off is that control of the principal is handed over to the platform.
Even if the underlying ETH is ultimately deposited into Ethereum's deposit contract, users still face the platform’s credit risk, internal control risks, and potential withdrawal restrictions during extreme market conditions.
2. Liquid Staking: LST Protocols Such as Lido
With liquid staking protocols such as Lido, users deposit ETH and receive an LST representing their staking position, such as stETH. The token can then be used in DeFi or sold on the secondary market.
The main advantage is that liquid staking lowers the participation threshold while allowing users to continue using their capital across DeFi and other applications.
However, it also adds more layers of risk. Holding an LST does not mean controlling a validator’s withdrawal credentials. Instead, the user holds a tokenized claim on pooled staked assets managed through smart contracts and protocol rules.
In addition to the underlying risk of validator penalties, users assume smart contract risk, governance risk, potential secondary-market discounts or depegging, and liquidity pressure during periods of heavy redemptions. LSTs have traded at discounts to the value of their underlying assets on multiple occasions.
3. Non-Custodial Native Staking: User-Controlled Withdrawals and Professional Validator Operations
Non-custodial native staking is designed to address the tension between asset sovereignty and the operational burden of running a validator.
Users participate through Ethereum’s native staking mechanism and retain control of their withdrawal address, while a professional service provider operates, monitors, and maintains the validator.
This model avoids the depegging risk associated with liquid staking tokens and reduces the additional smart contract and governance risks introduced by third-party staking pools.
4. Solo Staking
Users with sufficient ETH and technical expertise can also run their own validators.
Under this model, the user manages the signing key, withdrawal credentials, and validator hardware without depending on a third-party node operator.
But this does not necessarily make solo staking the lowest-risk option. Hardware failures, network outages, client misconfiguration, poor key backups, and duplicate signing can all lead to lost rewards or slashing.
For users without long-term validator-operation experience, the technical and maintenance burden can therefore be a significant barrier.
Taken together, these models draw their security boundaries in different places.
Different staking models serve users with different amounts of capital and different risk preferences. For users who prioritize direct control over their principal but do not want to maintain validator infrastructure themselves, non-custodial native staking is often the closest fit.
This is also the core design principle behind imToken's non-custodial ETH staking solution: users retain control of the withdrawal address, while professional infrastructure providers handle validator operations and maintain availability.
In practice, imToken provides two validator modes for different approaches to fund management:
- Compounding Validators: Eligible staking rewards remain in the validator balance and are automatically restaked, improving long-term compounding efficiency. A single validator can have an effective balance of up to 2,048 ETH.
- Auto-Withdrawal Validators: The validator’s effective balance remains at 32 ETH. Consensus-layer rewards above that amount are periodically and automatically transferred by the Ethereum protocol to the user’s withdrawal address, with no manual claim required. This mode is better suited to users who want regular access to their staking rewards.
Choosing a non-custodial model does not mean users can ignore security responsibilities.
Because users retain ultimate control over their assets, they must securely back up their wallet seed phrase offline, carefully verify the withdrawal address before staking, and understand Ethereum’s exit queue, service fees, and potential protocol-level penalties.
Final Thoughts
As more ETH enters long-term staking, holders need to consider more than how much their assets can earn.
The more important question may be this: if an external service fails under extreme conditions, do I still retain the ultimate authority to recover my principal?
Retain ultimate control of your assets while leaving validator operations to professionals.
That may be the most important security value of non-custodial native staking.