Academy glossaryDecision concept

Custody

Custody is not only a storage choice. It defines who can lose the assets, freeze access, misuse deposits, or make recovery impossible. The gap between self-custody and delegated custody changes every later safety decision.

You will see this in
self-custody wallet
exchange-held funds
multisig treasury control
How to use this page
Read the definition, then jump straight to the one decision this term should change.
Use the lesson and checklist blocks below when the term affects real execution behavior.
Treat the examples as product anchors so the term becomes easier to recognize under pressure.

Start with the term

Definition

Who really controls the keys, withdrawals, and final authority over your assets, even when the balance on screen still looks like yours.

Anchor 1
self-custody wallet
Anchor 2
exchange-held funds
Anchor 3
multisig treasury control
Quick custody check
I know who can actually sign or withdraw the assets.
I know whether I hold direct control or only a platform claim.
If something fails, I know whether recovery depends on me or on someone else.

How to spot and use it

Use these as the fast operational read: where the term first appears, what to watch for, and what rule should change your next move.

Spot first
self-custody wallet
Watch for
Any workflow where you cannot clearly name who controls the assets right now.
Rule
If another entity controls withdrawals or keys, treat that as part of the risk model, not as a boring platform detail.
Core lesson

Learn it properly

Work through the main concept first, then move into applied judgment and next actions.

What custody really decides

Custody is not a branding choice between 'wallet' and 'platform.' It is the answer to who can move the assets, who can lose them, and who gets to decide what happens when something goes wrong.

Self-custody means you control the keys and the recovery burden.
Delegated custody means someone else controls settlement and failure handling even if your balance looks familiar.
Claims on assets are not the same thing as direct onchain control.
Every later safety habit changes depending on whether you hold the authority or only trust someone else to hold it well.
Custody changes the shape of risk before the first trade, bridge, or approval ever happens.

Where users get custody wrong

The most expensive custody mistake is assuming visibility equals control. A number on a screen can look like ownership long after the operational authority sits somewhere else.

A custodial platform can still show balances while the underlying assets are mismanaged or unavailable.
Support access and account familiarity do not replace direct settlement rights.
Multisig or treasury control is still custody. It is just organizational custody instead of personal custody.
Good custody thinking asks who can sign, who can withdraw, and who can block recovery under stress.
If you do not know who holds the authority, you do not yet know what kind of asset risk you really have.
Real cases

What actually happened

These are public cases and repeated real-world patterns turned into teachable stories. Use them to see how small shortcuts become expensive outcomes in real product flows.

Public source-backed
Read the story first, then notice the exact decision that made the damage possible.
Case study

Customers saw balances. The exchange still owed them $8.7B

Loss: $8.7B owed to customers
Situation

After FTX collapsed, the bankruptcy team said the exchange owed customers about $8.7 billion. Roughly $6.4 billion of that was described as fiat and stablecoin that had been misappropriated.

Why this case matters

One real-world failure usually teaches faster than ten abstract warnings.

What they assumed

If the interface shows the assets and withdrawals usually work, the customer still effectively controls the money in any meaningful sense.

Red flag you would have seen in the UI

A balance view that feels like ownership even though the real custody and settlement authority sit with the platform. In product terms, visibility of funds is not the same thing as control of funds.

You would have seen this on

These are the exact product moments where this kind of mistake usually first looks harmless.

Wallet promptStatus
What went wrong
1
Customers treated an account balance like direct control over assets.
2
The platform commingled and misused customer deposits behind the scenes.
3
When trust failed, users discovered they held claims, not immediate authority.
4
The shortfall reached about $8.7 billion according to the bankruptcy team's findings.
Core lesson

Custody risk is often invisible right up until the moment users discover that a displayed balance is not the same thing as operational control.

What they should have done instead

Treat custody as a first-order question. If another party controls the assets, be honest that you hold platform risk in addition to market risk.

Core points

Why it changes the decision

It decides who actually has the power to lose, freeze, misuse, or delay the assets.
It changes the meaning of recovery, support, and trust before any route is even selected.
It is one of the biggest differences between wallet-first DeFi behavior and exchange-like assumptions users bring from Web2.
A user who does not understand custody often thinks they own an asset operationally when they only have a claim on it.
Use after the lesson

Before you sign or confirm

This section should help in the moment of risk. Keep one question in mind: what should I check right now before giving authority or sending the route forward?

Check now
Do not think in abstract principles here. Think in checks you can do on this screen before moving forward.
Do now
Ask who actually controls the keys, withdrawals, or signing surface.
Separate market risk from custody risk in your head before trusting the setup.
Treat delegated control as a real risk layer, not a UX convenience detail.
Do not continue if
Do not assume the interface tells you who really controls the money.
Do not equate platform responsiveness with safe custody architecture.
Do not ignore custody risk just because the product flow feels smooth.
Red flag if this feels routine
If this step feels like harmless friction, that is already the red flag.
1
Any workflow where you cannot clearly name who controls the assets right now.
2
Any asset view that feels like ownership while another party controls withdrawal rights.
3
Any organization or platform where key authority is opaque until something breaks.
Before first serious use
If these checks are not clear yet, you are not in a good position to rely on speed or instinct.

Quick custody check

1
I know who can actually sign or withdraw the assets.
2
I know whether I hold direct control or only a platform claim.
3
If something fails, I know whether recovery depends on me or on someone else.
4
I am pricing custody risk separately from token or market risk.
Use after the lesson

Decision flow

Do not use this like a reading section. Use it as the order of operations when the screen is asking for authority or final confirmation.

How to think through it

1
Step 1

Start with who controls the keys

Before asking whether the trade or platform looks attractive, ask who actually holds the authority over the assets right now.

2
Step 2

Translate that into failure reality

If another party controls settlement or withdrawals, your failure mode is not only market loss. It is also their solvency, honesty, and operational behavior.

3
Step 3

Decide whether the custody trade-off is worth it

Some delegated custody may be acceptable, but only if you are explicit that you are trusting another layer, not pretending it disappeared.

Signals to notice

1
You can see balances but cannot verify who can move the funds

That usually means you understand the interface better than the custody reality.

2
Recovery depends mostly on another company's promises

That is a custody signal, not only a support signal.

3
You treat centralized balances and self-custodied balances as the same kind of exposure

That flattens two very different authority models into one dangerous assumption.

Rules

Decision rules

If another entity controls withdrawals or keys, treat that as part of the risk model, not as a boring platform detail.
If recovery depends on someone else's solvency, honesty, or operations, you are carrying custody risk even if the asset price is stable.
If you cannot clearly say who can sign or move the funds, stop calling the setup safe by default.
The more distant you are from the real authority, the more careful you should be about trust concentration.
Avoidable errors

Common mistakes

Confusing balance visibility with direct control.
Treating platform familiarity as proof that custody is safe.
Ignoring who can really sign, freeze, or withdraw under stress.
Thinking custody only matters for centralized exchanges and not for treasury or multisig setups.
Practice

Short scenarios

Use quick situations like these to test whether the concept would hold up in a real product flow.

Balance visible, authority elsewhere

A platform clearly shows your funds and past withdrawals worked fine, so the setup feels operationally identical to direct control.
Treat that as a custody illusion until you can answer who holds the withdrawal authority and what happens if the platform itself fails.

Treasury multisig confidence

A team treasury uses a multisig and everyone assumes that means custody risk is solved by default.
A multisig improves structure, but it is still custody. The real question is who signs, what they see, and how signer behavior fails under pressure.
Continue learning

Keep building the path

Once the core lesson is clear, use these paths to widen the mental model or go deeper where the concept matters most.

Continue learning

Related Academy paths

Once the core lesson is clear, use these paths to widen the mental model or go deeper where the concept matters most.

    Custody | ZeroLyx Academy Glossary