EST · MMXXVI
Home/Insights/Regulatory/Custody Insurance and Cyber Cover: Reviewing the Policy
Licensing & Registration

Custody Insurance and Cyber Cover: Reviewing the Policy

Custody Insurance and Cyber Cover: Reviewing the Policy. Independent digital-asset law for exchanges, issuers and funds. Cross-border scope, fixed-fee engagemen

A custodian preparing for a VASP licence in a major hub discovers that the insurer's policy excludes "losses arising from private-key compromise" – the very risk the regulator expects the business to cover. The disconnect is not unusual. Standard crime and cyber policies were written for fiat financial institutions; digital-asset custody sits at the intersection of regulated safeguarding, cryptographic control and counterparty risk, and the insurance market has not yet caught up uniformly across all these dimensions.

Custody insurance and cyber cover for digital-asset businesses must align with three moving parts: the regulated custody perimeter (what counts as "custody" under the applicable VASP or financial-services regime), the technical control model (who holds keys, whether self-custody or third-party, hot or cold storage), and the policy language itself – which often excludes precisely the operational risks that regulators demand a custodian insure. This analysis examines what crime, specie and cyber policies actually cover for digital-asset custodians, the common exclusions that create gaps between regulatory expectations and underwritten risk, and how to structure cover that satisfies both the regulator and the underwriter. We walk through the decision points for an exchange or fund manager seeking authorisation in a regime that mandates insurance as a condition of the custody licence, and we address the question businesses routinely ask after the first policy draft arrives: does this policy actually protect us if something goes wrong?

What is custody under the regulatory framework?

The regulated custody perimeter is defined by control over client assets, not by the label on the service or the custody technology in use. Under MiCA, custody and administration of crypto-assets on behalf of clients is a standalone regulated activity; a CASP providing this service must segregate client holdings, maintain adequate financial resources and, in many member states, hold insurance or an equivalent guarantee covering operational and security risk. Under VARA in Dubai, custody is one of the core Virtual Asset activities requiring a licence; the rulebook sets out detailed segregation, safeguarding and financial-resilience expectations. The same logic applies across Singapore (MAS), Hong Kong (SFC), Switzerland (FINMA), the BVI and the Cayman Islands: if you hold, administer or have the technical ability to dispose of a client's virtual assets, you are providing custody, and a raft of prudential requirements – capital, governance, audit, and often insurance – follows.

The regulator's definition of custody hinges on two questions. First, does the business have de facto control – the technical ability to sign a transaction moving the client's tokens? If the answer is yes, the activity is custody, regardless of whether the business advertises itself as a "wallet provider", an "administrator" or a "non-custodial platform". Second, does the client rely on the business to safeguard the asset? If users deposit funds into an omnibus wallet controlled by the platform and withdraw on instruction, the platform is a custodian. Self-custody models – where the user alone holds the private key and the platform never gains signing authority – generally fall outside the custody perimeter, but even here the line can blur: a platform that offers backup-key recovery or holds a shard in a multi-signature scheme may be deemed to provide custody or custody-adjacent services in certain regimes.

This matters for insurance because the regulated obligation to insure tracks the custody perimeter, not the technology stack. If MiCA, VARA or MAS says "custodians must maintain insurance covering loss of client crypto-assets", the regulator expects that policy to respond when client tokens are misappropriated, whether the loss arises from an external hack, an insider breach or a key-management failure. The disconnect arises when the business presents a standard cyber-liability or crime policy to the regulator, and the policy excludes "loss of intangible property" or "loss of bearer instruments" – language drafted for a world in which custody meant holding securities in a vault, not managing cryptographic keys on a server.

Crime policies and the intangible property exclusion

Traditional crime insurance (often written as a financial-institution bond) covers employee dishonesty, forgery, theft of physical property and, in some forms, loss of securities in transit or from a depository. The policy defines "property" narrowly: cash, negotiable instruments, physical certificates. Digital assets frequently do not fit that definition, and many crime policies contain an explicit exclusion for "loss of data" or "loss of intangible property" – a category into which underwriters initially tried to place crypto-assets. The effect: a custodian suffers a six-figure loss when an employee extracts private keys and transfers client tokens to an external address, and the crime insurer denies the claim on the ground that the policy covers theft of "property", not theft of data or digital entries.

Specialty insurers now offer digital-asset crime or specie policies that explicitly include crypto-assets in the definition of covered property. These policies recognise that a private key controlling a token balance is not "data" in the general sense; it is the functional equivalent of a bearer instrument. The better digital-asset crime policies define "crypto-asset" as a unit of value recorded on a distributed ledger and list the covered activities: theft by employee dishonesty, external hacking, social engineering (where an employee is tricked into authorising a fraudulent transaction), and loss from a third-party custodian or sub-custodian. The policy will specify the insured perimeter – for example, tokens held in the custodian's hot wallets, cold-storage vaults or with a named sub-custodian – and any retention or deductible.

Common exclusions remain. Many policies exclude losses arising from "voluntary transfer" – meaning that if the business (or an authorised employee) deliberately moved the tokens pursuant to a fraudulent instruction, the claim may be denied unless the policy includes a social-engineering or funds-transfer-fraud extension. Some exclude losses from smart-contract vulnerabilities or protocol failures – the policy covers theft of the private key, not a loss resulting from a flaw in the DeFi protocol the custodian integrated. Others exclude regulatory fines, penalties or "loss of business income" – so if a hack forces the platform offline for a week and clients withdraw en masse, the policy reimburses the stolen crypto but not the consequential revenue loss or the regulatory penalty for breaching uptime commitments.

In our licensing practice, we regularly see applications stall at the insurance stage because the business submitted a standard crime policy with an intangible-property exclusion still in place. The regulator will reject that policy as non-compliant, and the applicant must return to the broker with a clear brief: the policy must explicitly cover crypto-assets as property, must respond to both internal and external theft, and must align the insured perimeter with the regulated custody perimeter – typically the full balance of client assets the business holds or controls. Getting that alignment right is not a one-way conversation with the insurer; it requires a detailed factual presentation of the custody model (key-management architecture, signing authority, sub-custodian arrangements, multi-signature thresholds) and, often, a legal opinion confirming that the regulated activity the business will perform is "custody" as defined in the target jurisdiction.

Cyber liability versus first-party digital-asset loss

A cyber-liability policy typically covers third-party liability arising from a data breach – defence costs, regulatory fines under privacy laws, customer notification expenses, credit-monitoring services – and may include first-party coverage for business interruption and data restoration. What it generally does not cover is the direct theft of digital assets from a wallet under the insured's control. Cyber policies treat crypto-assets as "money or securities" (often subject to a sub-limit or complete exclusion) or as "intangible property" (excluded). The result: a custodian suffers a breach in which attackers exfiltrate client credentials and drain hot wallets, and the cyber policy pays for forensic investigation, legal defence and notification letters, but not for the value of the stolen tokens.

Some specialist cyber policies for digital-asset businesses now include a first-party "crypto-asset loss" extension, bridging the gap between the traditional cyber product and the crime/specie policy. These extensions cover theft of tokens resulting from a cyberattack – for example, a ransomware incident that compromises key-management infrastructure or a phishing campaign that tricks an operator into signing malicious transactions. The cover sits alongside the liability and breach-response elements, giving the custodian a single policy that responds to both the theft of the asset and the regulatory and reputational fallout. The limits are typically lower than standalone crime cover (a cyber policy might offer $5–10 million in crypto-asset loss cover as a sub-limit within a $25 million aggregate), so businesses operating at scale often layer a dedicated specie policy on top.

Key questions when reviewing a cyber policy for custody adequacy: Does the policy define "computer fraud" or "funds-transfer fraud" to include transactions in crypto-assets, or is it limited to fiat wire transfers? Does the business-interruption cover respond to downtime caused by a protocol failure or a denial-of-service attack on the blockchain infrastructure the custodian relies on, or only to direct attacks on the custodian's own systems? Are losses from a sub-custodian breach covered – in other words, if the business uses a third-party cold-storage provider and that provider is hacked, does the cyber policy respond, or must the business pursue a separate claim against the sub-custodian's insurer? And does the policy contain a "war and terrorism" or "nation-state actor" exclusion – relevant for digital-asset businesses, given that a significant proportion of large-scale hacks involve state-sponsored groups?

Aligning cover with the regulated custody perimeter

The central tension in custody insurance is that the regulator's expectations and the insurer's standard exclusions pull in opposite directions. The regulator says: you must hold insurance covering the full balance of client crypto-assets under your custody. The insurer says: we will cover crime and cyber events up to a limit, but we exclude losses from smart-contract bugs, protocol failures, losses while tokens are in transit between wallets, and any loss where the keys were held by a sub-custodian unless you separately insured that arrangement. Squaring that circle requires a three-step process: map the custody model, quantify the risk at each stage, and structure cover that either eliminates the gaps or makes them transparent to the regulator.

Step one: map the custody model. Identify every point at which the business (or a third party on its behalf) holds, administers or has signing authority over client tokens. This includes hot wallets (used for day-to-day transactions), warm wallets (intermediate storage with some automated access), cold wallets (offline or air-gapped), and any tokens held with a sub-custodian or on a third-party exchange pending settlement. For each category, document who holds the private keys, whether multi-signature or threshold-signature schemes are in place, and the maximum balance expected at any given time. This is the insured perimeter – the inventory of assets the policy must cover.

Step two: identify the risk layers. The largest category of loss historically is external hacking (an attacker compromises the platform and extracts keys or tricks the system into signing unauthorised transactions). Second is insider theft (an employee or contractor with key access moves tokens to an external address). Third is social engineering (a sophisticated fraud in which an attacker impersonates a legitimate counterparty and induces the business to transfer funds). Fourth is sub-custodian failure (the third-party cold-storage provider suffers a breach or becomes insolvent). Fifth is smart-contract or protocol failure (a bug in a staking contract or a flaw in the underlying blockchain leads to permanent loss). Different policies cover different subsets of these risks, and the regulator will want to see that the first four are insured; the fifth is often left uninsured or is addressed through contract warranties and technical audit, not through an insurance policy.

Step three: structure the policy stack. For a mid-sized custodian, this often means a layered approach. A primary crime/specie policy covering employee dishonesty, external theft and social engineering, with a limit set at or above the peak client-asset balance. A cyber-liability policy with a first-party crypto-asset-loss extension, providing excess cover above the crime policy and covering business interruption and breach response. And, if the business relies on a sub-custodian, either a representation from the sub-custodian that it holds its own insurance naming the business as a loss payee, or a separate contingent-cover endorsement on the primary policy. The total limit must meet or exceed the regulatory minimum – which varies widely: some regimes set no specific floor but require "adequate" cover relative to the business's risk profile, while others specify a percentage of assets under custody or a fixed monetary threshold.

In a recent licensing engagement, an exchange applying for a MiCA CASP authorisation in a member state presented a $10 million crime policy and a $15 million cyber policy, both from reputable carriers. The national competent authority rejected the package on the ground that the crime policy excluded "loss of tokens held with a third-party custodian" and the business used a sub-custodian for cold storage holding 60 per cent of client balances. We restructured the cover: the business negotiated a policy amendment removing the third-party exclusion in exchange for a higher premium, obtained a certificate from the sub-custodian's insurer confirming that the business was named as an additional insured, and added a $5 million excess layer above the primary policies to lift the aggregate limit to a level the regulator deemed proportionate to the projected asset base. The revised package was approved, and the authorisation moved forward.

Common policy exclusions and how they create gaps

Understanding what a policy does not cover is as important as understanding what it does. The exclusions most likely to create a gap between the business's risk exposure and the insured cover are the following:

Voluntary-transfer exclusions. Many policies exclude losses where the insured (or an authorised employee) voluntarily transferred the property. On its face this makes sense – the policy covers theft, not legitimate transactions. But in the digital-asset context, "voluntary transfer" becomes ambiguous: if an attacker uses a phishing campaign to obtain credentials and the business's system, acting on those credentials, executes a transaction, is that voluntary? Some courts and insurers have said yes, denying the claim. Better policies replace the blanket voluntary-transfer exclusion with a narrower carve-out and add a social-engineering or fraudulent-instruction endorsement, clarifying that a transfer induced by fraud is covered even if the business's system executed it.

War, terrorism and nation-state exclusions. Standard cyber policies often exclude losses caused by "war, invasion, hostilities, terrorism or action by a foreign government". Given that a material proportion of large digital-asset hacks involve groups operating from or with the tacit support of certain states, this exclusion can be fatal to a claim. Some specialist insurers now offer a limited buy-back: the war exclusion remains but is narrowed to acts that meet a legal threshold of armed conflict, and the nation-state exclusion is amended to apply only where a government itself (not a private group within that state) directly ordered the attack. Negotiating these terms at the outset is essential; after a loss, the insurer will point to the exclusion and the business will have no recourse.

Loss of use or loss of market value. Most policies cover the "direct physical loss" of the crypto-asset – meaning the value at the time of theft – but exclude consequential losses such as the opportunity cost of holding a token that subsequently appreciated, or the loss of staking rewards the business would have earned. This is standard in property insurance. The business should understand that if 100 ETH are stolen on the day they are worth $200,000, the policy pays $200,000 (minus any deductible), even if those tokens would have been worth $300,000 by the time the claim settles.

Smart-contract and protocol failures. If a token is lost because of a bug in a staking smart contract or a consensus failure in the underlying blockchain, most policies will not respond – the loss did not result from theft or a cyberattack on the custodian. Some insurers offer smart-contract insurance as a separate product, covering losses from code bugs in specified contracts. This is common for DeFi protocols but less common for custodians. The regulatory expectation is usually that smart-contract risk is managed through audit and testing, not through insurance, so the absence of this cover is typically not an obstacle to authorisation – but the business should be clear internally that this risk sits with the business, not with the insurer.

Fines, penalties and punitive damages. Most crime and cyber policies in the US exclude fines and penalties imposed by government agencies (though they may cover defence costs and compensatory damages). In the EU and other jurisdictions, coverage for regulatory fines is often available and in some cases is a regulatory expectation for certain licensed activities. When structuring cover for a cross-border custodian, the business should check the governing law of the policy and confirm whether regulatory-penalty cover is included or available by endorsement. In some member states under MiCA, the regulator may require evidence that fines for operational failures (for example, a breach of segregation rules) are covered, not just the direct loss of assets.

Sub-custodian arrangements and the chain of cover

Many custodians use a third-party sub-custodian for cold storage, either a specialist digital-asset custodian or a qualified custodian under applicable securities or banking rules. This introduces a second layer of risk: the sub-custodian may itself suffer a breach, or it may become insolvent and the client tokens become caught in an insolvency proceeding. The primary custodian's insurance may or may not cover losses arising from the sub-custodian's failure, depending on how the policy defines the insured perimeter.

Best practice is to ensure that one of three conditions is met. First, the primary custodian's policy explicitly covers "tokens held by a sub-custodian on the insured's behalf" and names the sub-custodian (or a class of sub-custodians meeting certain criteria) in a schedule. Second, the sub-custodian holds its own insurance and names the primary custodian as a loss payee or an additional insured, so that the primary custodian can claim directly against the sub-custodian's insurer if a loss occurs. Third, the custody agreement with the sub-custodian includes a representation that the sub-custodian maintains insurance meeting a stated standard and an obligation to provide a certificate of insurance annually. In practice, many large qualified custodians hold substantial crime and E&O cover and will provide a loss-payee endorsement as a matter of course; smaller or offshore sub-custodians may not, and in that case the primary custodian must either self-insure that risk or find a different provider.

Regulators increasingly scrutinise sub-custodian insurance in the authorisation process. Under MiCA, if a CASP delegates custody to a third party, the CASP remains liable to the client for any loss, and the regulator will expect either that the CASP's insurance covers the full chain or that the sub-custodian's insurance is of equivalent quality and that the CASP can demonstrate it has verified this. A licensing application that says "we use [Sub-Custodian X] and assume they are insured" without documentary evidence will not satisfy the regulator. We routinely request and review the sub-custodian's certificate of insurance, confirm that the limits and exclusions align with the primary policy, and include a summary of the sub-custodian cover in the licensing submission.

How insurance premiums are priced for digital-asset custody

Pricing for custody insurance varies with the risk profile of the business, the coverage limit, the deductible and the underwriter's assessment of the controls in place. Premium rates for digital-asset crime/specie cover are significantly higher than for traditional financial-institution bonds – historically in the range of 2 to 5 per cent of the sum insured for a newly licensed custodian, falling to 1 to 2 per cent for an established business with a strong control environment and no claims history. Cyber-liability premiums add another layer, typically 0.5 to 1 per cent of the limit. For a business seeking $20 million in combined crime and cyber cover, the annual premium might fall in the low six figures initially, declining as the business matures and demonstrates operational resilience.

Underwriters assess several factors. Key-management architecture is paramount: multi-signature wallets, hardware security modules, geographically distributed key shards and time-locked withdrawals all reduce the underwriter's perceived risk and can drive the premium down. A business using a single hot wallet with keys stored on a cloud server will pay a materially higher rate than one using a multi-layered cold-storage model with institutional-grade signing infrastructure. Audit and penetration testing also matter: a business that has passed a SOC 2 Type II audit and undergoes quarterly penetration tests will find more insurers willing to quote and at better rates. Third, the regulatory environment – a custodian licensed under MiCA or by MAS benefits from the halo effect of a rigorous authorisation process, whereas a custodian operating without a licence or in a jurisdiction with weak supervision may struggle to obtain cover at any price.

The deductible (or retention) for digital-asset crime policies is typically set at a level that aligns with the business's risk appetite and capital base – commonly between $100,000 and $1 million. A higher deductible reduces the premium but leaves the business exposed to smaller losses. Regulators do not usually specify a maximum deductible, but they do expect that the business has sufficient liquid capital to absorb the deductible in a loss scenario without jeopardising its ability to meet client obligations. This ties back to the broader financial-resilience requirements: a custodian must hold a minimum level of own funds (capital) and insurance; the two are complementary, not substitutes.

Timing insurance in the licensing process

A common mistake is to apply for a VASP or CASP licence without securing at least an indicative insurance quote. Many applicants assume they can "sort out insurance later" once the licence is granted, only to discover that obtaining compliant cover takes weeks and that the regulator will not issue the final authorisation until the policy is in place and evidenced. The better sequence is to approach brokers and underwriters in parallel with the licensing preparation, provide them with a detailed description of the custody model, and obtain a binding quote or a letter of intent that can be submitted as part of the application. The letter should state the proposed coverage (crime and cyber), the limit, the deductible, the premium and any material conditions, and it should confirm that the insurer is prepared to issue the policy upon the granting of the licence.

In practice, insurers are willing to quote for a business that does not yet hold a licence but is in the process of applying. They price the risk on the assumption that the business will meet the regulatory requirements – so the quote is conditional on the licence being granted and on the business implementing the controls it described in the underwriting submission. Once the regulator issues the authorisation, the insurer binds the cover and the policy becomes effective, typically on the same day or shortly thereafter. This allows the licensing process and the insurance process to run in parallel, and it avoids the situation where the regulator grants conditional approval but the business cannot commence operations because it cannot obtain the required insurance.

For businesses seeking authorisation in multiple jurisdictions, the insurance requirement may differ across those jurisdictions. MiCA will require CASP-level cover; VARA in Dubai may specify a minimum limit based on the volume of assets under custody; Singapore may accept a lower limit for a standard payment institution providing custody as an ancillary service. The business should map these requirements early and determine whether a single global policy can be structured to meet all of them, or whether separate policies (or regional endorsements) are needed. Global insurers and specialist brokers can usually accommodate multi-jurisdiction programmes, but the coordination adds time, so starting the process early in the licensing timeline is essential.

The regulator's perspective: what counts as adequate?

When a regulator reviews a custody-insurance package, the regulator is asking two questions. First, does the policy actually cover the risk the regulation is designed to mitigate – namely, the risk that client crypto-assets under the custodian's control are lost, stolen or misappropriated? Second, is the coverage limit proportionate to the business's scale and the potential exposure? A $5 million crime policy might be adequate for a fund manager providing custody to a handful of institutional clients with a peak balance of $10 million, but it is plainly inadequate for an exchange holding $500 million in client deposits.

Regulators do not prescribe the exact policy terms – that remains a commercial negotiation between the business and the insurer – but they do expect certain baseline features. The policy must define "crypto-assets" or "virtual assets" clearly and confirm that they are covered property. It must cover both internal and external causes of loss (employee dishonesty and hacking). It must respond to loss of client assets, not just the custodian's proprietary holdings. And the business must demonstrate that it has read the policy, understands the exclusions, and has a plan for managing the risks that fall outside the insured perimeter (for example, smart-contract bugs, protocol failures, or losses during a brief window when tokens are in transit between wallets).

In member states implementing MiCA, the national competent authority will scrutinise the policy as part of the CASP authorisation process. The business submits the policy wording (or the binding quote), and the regulator's legal and prudential teams review it. If the policy contains a problematic exclusion – for instance, excluding "loss arising from a failure of the insured's computer systems" – the regulator will flag it and ask how the business intends to cover that risk. The business must then either negotiate an amendment to the policy, obtain a separate endorsement, or provide a written explanation (often in the form of a legal opinion or a technical memo) demonstrating that the excluded risk is mitigated through other controls and is not material to the overall risk profile. This back-and-forth can add weeks to the authorisation timeline, which is why experienced applicants vet the policy language in detail before submitting it.

Decision matrix: structuring cover by business profile

The appropriate insurance structure depends on the business's operating model, the volume of assets under custody, and the regulatory requirements in the target jurisdiction. Below is a decision matrix, expressed in prose, that maps common custodian profiles to the typical policy stack and indicative limits:

Profile A: Small fund manager (assets under custody $5–20 million). The core risk is internal – a rogue employee or a compromised administrator account. Recommended cover: a $5–10 million crime/specie policy covering employee dishonesty, external theft and social engineering, with a $100,000 deductible, and a $10 million cyber-liability policy with a first-party crypto-asset-loss extension. Total cost typically in the range of $100,000–$150,000 annually. If the fund uses a qualified sub-custodian (for instance, a bank or a licensed digital-asset custodian), confirm that the sub-custodian's insurance names the manager as a loss payee, which may allow a reduction in the primary policy limit.

Profile B: Mid-sized exchange (peak client deposits $50–200 million). The risk is both internal and external, with a higher probability of a targeted cyberattack. Recommended cover: a $25–50 million crime/specie policy as the primary layer, excess cyber cover of $25–50 million (for a total limit of $50–100 million), and a separate $10 million policy for business interruption and regulatory defence costs. Use of multi-signature wallets and institutional custody for the bulk of cold storage can reduce premiums. Sub-custodian arrangements should be explicitly covered in the primary policy or by endorsement. Total annual premium typically in the mid-six figures, declining after the first policy year if no claims are made.

Profile C: Large custodian or institutional prime broker (assets $200 million+). The risk is systemic: a successful attack could result in a loss exceeding $100 million. Recommended cover: a $100 million+ layered programme, with a primary crime policy of $25–50 million and multiple excess layers placed with different insurers to spread the risk. Cyber cover and E&O (errors and omissions) cover should be layered on top, with particular attention to the definition of "custody" and the treatment of staking, lending and other ancillary activities. The business should also consider purchasing separate cover for regulatory penalties and for defence costs in the event of a mass-tort or class-action suit following a breach. Total annual premium in the low seven figures. This level of cover is typically required for a custodian seeking a banking or securities-dealer licence in a Tier 1 jurisdiction.

Profile D: Non-custodial platform (self-custody model, but providing backup-key recovery). Technically the platform does not hold client funds day-to-day, but it does hold encrypted key shards for account recovery, which may bring it within the custody perimeter depending on the jurisdiction. Recommended cover: a smaller crime policy ($5–10 million) limited to losses arising from compromise of the recovery infrastructure, and a robust cyber-liability policy covering privacy breaches and business interruption. The platform should obtain a legal opinion clarifying whether the backup service is regulated custody in the target jurisdiction; if it is, the policy must be structured accordingly. If it is not, the regulatory burden is lighter, but the business still faces reputational and operational risk if the recovery system is breached.

Common mistakes and how to avoid them

In our licensing and compliance practice, we see the same missteps recur. First, businesses wait too long to engage an insurance broker. Obtaining a binding quote for digital-asset crime and cyber cover is not instantaneous; underwriters require detailed information about the technology stack, the control environment, the team's experience and the regulatory status, and they may request a site visit or a technical audit before committing to cover. Starting the insurance process only after the licence application is submitted compresses the timeline and may force the business to accept a less-than-ideal policy simply to meet a regulatory deadline.

Second, businesses assume that a generic cyber policy is sufficient and do not read the exclusions. A policy that covers "computer fraud" in the context of fiat wire transfers does not necessarily cover theft of crypto-assets. The business should insist on policy wording that explicitly includes digital assets in the definition of covered property and that contains a clear statement of what is and is not covered. If the policy is ambiguous, the business should request an endorsement or a clarification letter from the insurer before submitting it to the regulator.

Third, businesses fail to coordinate the insurance requirement with the minimum-capital requirement. Some regimes allow the business to reduce its capital floor if it holds insurance above a certain threshold; others treat insurance and capital as independent requirements. The business should understand how these two elements interact in the target jurisdiction and should structure the capital and insurance package together, not in isolation. A related mistake is assuming that insurance is a substitute for good operational controls. No policy will respond if the business suffers a loss because it ignored basic security hygiene – for example, storing private keys in plaintext on a server, or failing to implement multi-signature controls despite promising to do so in the licence application. Insurers conduct post-loss investigations, and if they discover that the business misrepresented its controls in the underwriting process or failed to maintain them, the claim will be denied.

Fourth, businesses overlook the claims process. A policy is only as good as the insurer's willingness and ability to pay. The business should check the insurer's financial rating (A.M. Best, S&P, Moody's) and should understand how claims are filed, what documentation is required, and the typical settlement timeline. In a digital-asset theft scenario, time is critical – the business needs to freeze accounts, notify law enforcement and possibly pursue on-chain recovery – and a slow or adversarial claims process can compound the damage. Some brokers recommend placing cover with multiple insurers in layers precisely to avoid concentration risk: if one insurer disputes the claim, the others may still pay their share.

Cross-border considerations: multi-jurisdiction programmes

For a business operating across borders – a common reality in digital assets, where users, servers and bank accounts may sit in different jurisdictions – the insurance challenge is to structure a programme that meets the requirements of each regulator without creating gaps or overlaps in coverage. A global custody platform might hold a CASP licence in a member state, a VARA licence in Dubai and a Payment Services Act licence in Singapore. Each regulator expects evidence of insurance; each may have slightly different requirements as to limits, exclusions and the definition of the insured perimeter.

The standard approach is to obtain a master policy written on a global basis, with jurisdiction-specific endorsements confirming that the policy complies with local regulatory requirements. The master policy defines the core cover (crime, cyber, limits, exclusions), and the endorsements adjust specific terms where needed – for example, one endorsement might confirm that regulatory fines in the EU are covered, while another confirms that the policy responds to claims brought under Singapore law. The business provides each regulator with a certificate of insurance (often countersigned by the insurer) stating that the cover meets or exceeds the local requirement. This avoids the need to purchase separate policies in each jurisdiction, which would be both more expensive and operationally complex (managing multiple renewal dates, deductibles and claims processes).

A challenge arises where different jurisdictions define custody differently. Under MiCA, staking-as-a-service may be treated as custody if the business controls the staked tokens; under another regime it may be classified as an advisory or execution service, outside the custody perimeter. The insurance policy should ideally be drafted broadly enough to cover all the activities the business provides across all its licences, with the understanding that the policy responds based on the facts of the loss, not on the regulatory label applied to the activity. In practice this means defining "custody" in the policy by reference to "control over client crypto-assets" rather than by reference to any single jurisdiction's legal definition.

For a business expanding into a new jurisdiction after already holding insurance, the process is simpler: the business notifies the insurer of the additional licence, provides updated information about the volume of assets and the new regulatory requirements, and requests an endorsement adding the new jurisdiction to the insured locations. The insurer may adjust the premium to reflect the increased exposure, but there is usually no need to renegotiate the entire policy. This modular approach – a core global policy with jurisdiction-by-jurisdiction endorsements – is how most multi-national digital-asset custodians structure their insurance, and it is a model regulators are increasingly familiar with and accept.

When the insurance requirement and the licensing timetable intersect, early coordination between legal, compliance, operations and the insurance broker is essential. A licensing application that says "insurance will be obtained prior to launch" without supporting detail is weaker than one that includes a binding quote or a letter of intent from an insurer, demonstrating that the business has already secured coverage in principle. For a comprehensive review of your custody model and the insurance package that will support your authorisation, contact OBOLUS at info@oboluslaw.com.

How OBOLUS reviews custody insurance packages

In a typical engagement, a business preparing for a CASP, VARA or MAS licence asks us to review the insurance package the broker has proposed. We start by mapping the business's custody model – the full inventory of client assets, the key-management architecture, the use of sub-custodians, the withdrawal and settlement flows – and we identify the regulatory custody perimeter in the target jurisdiction. We then review the policy wording line by line, checking the definition of covered property, the insuring clauses, the exclusions, the limits and the claims process. We flag any terms that conflict with the regulatory expectation or that leave a material risk uninsured, and we provide the broker with a redline or a list of required amendments.

Where the policy contains a problematic exclusion, we assess whether it can be negotiated out, whether a separate endorsement is available, or whether the risk is sufficiently remote that the regulator will accept a written explanation. For example, if the policy excludes "loss during transit between wallets", we might explain to the regulator that the business uses atomic swaps or time-locked scripts such that tokens are never in an uncontrolled state, and that the transit window is measured in seconds, not hours, reducing the exposure to a de minimis level. This kind of contextual explanation, backed by technical documentation, can satisfy a regulator even where the insurance does not cover every conceivable scenario.

We also coordinate with the broker to ensure that the policy is presented to the regulator in a format the regulator expects – typically a summary or certificate of insurance, the full policy wording, and a cover letter from the insurer confirming that the policy meets the local regulatory requirement. In some jurisdictions the regulator requires that the insurer holds a specified financial rating or is licensed to write insurance in that jurisdiction; we verify that the insurer meets these criteria before the policy is submitted. And we monitor the licensing process: if the regulator raises questions about the insurance, we liaise with the broker and the insurer to address them quickly, avoiding unnecessary delays in the authorisation timeline.

Beyond the initial licensing phase, we advise on the ongoing maintenance of the insurance – annual renewals, mid-term amendments to reflect changes in the business (for example, an increase in assets under custody or the addition of a new sub-custodian), and claims strategy in the event of a loss. Insurance is not a static document; it must evolve as the business scales and as the regulatory environment changes. A business that obtained a $10 million crime policy three years ago and now holds $100 million in client assets is materially underinsured, and the regulator will expect the policy to be adjusted. We work with clients to build a forward-looking insurance roadmap that aligns with their growth projections and their licensing plans, ensuring that they remain compliant in every jurisdiction where they hold or intend to hold a licence.

When a claim arises: navigating the process

If a loss occurs – for example, a cyberattack results in the theft of client crypto-assets – the business must act immediately on two parallel tracks: operational response (freezing accounts, notifying law enforcement, pursuing on-chain recovery, informing the regulator and affected clients) and insurance claim. The policy will contain a notice provision requiring the business to notify the insurer "as soon as practicable" or within a stated period (commonly 30 to 90 days) of discovering the loss. Failing to notify within that window can jeopardise the claim.

The business should notify the insurer in writing, providing a preliminary description of the loss (the tokens stolen, the approximate value, the circumstances) and reserving the right to supplement the claim as more information becomes available. The insurer will assign a claims adjuster and, in many cases, will appoint a forensic investigator to examine the cause of the loss and verify the business's account. This investigation may take weeks or months. The business should cooperate fully but should also engage its own forensic and legal advisers to ensure that the facts are documented accurately and that the business's interests are protected.

Common issues in the claims process include valuation disputes (the insurer may value the stolen tokens as of the date of theft, while the business argues for a later date when it discovered the loss and the tokens had appreciated), coverage disputes (the insurer argues that the loss falls within an exclusion, such as the voluntary-transfer exclusion, while the business argues that the transfer was fraudulent and thus covered), and documentation disputes (the insurer requests transaction logs, key-custody records and access to servers, and the business must balance cooperation with the need to preserve privilege and confidentiality). Having legal counsel involved from the outset of the claims process can help navigate these issues and can increase the likelihood of a favourable settlement.

In our disputes and recovery practice, we have represented custodians and exchanges in insurance claims following cyberattacks and internal fraud. The claims process is adversarial – the insurer's duty is to its shareholders, not to the insured, and the adjuster will look for any basis to deny or reduce the payout. A well-drafted policy, clear documentation of the loss, and early engagement of experienced coverage counsel are the keys to a successful claim. If the insurer denies the claim or offers a low settlement, the business may pursue coverage litigation or arbitration (depending on the dispute-resolution clause in the policy); these proceedings can take years, so many claims settle through negotiation before reaching that stage.

The reality is that insurance is as much about managing the relationship with the insurer as it is about the policy terms. A business that has maintained open communication with the insurer, provided regular updates on its risk profile and invited the insurer to participate in audits or risk reviews will generally find the claims process smoother than one that treated the policy as a box to tick and had no contact with the insurer until a loss occurred. If you are managing a custody operation and the operational or insurance picture is shifting, reach OBOLUS now at info@oboluslaw.com.

FAQ

How long does a crypto licence take to obtain?

The timeline varies by jurisdiction and by the completeness of the application. A straightforward VASP registration in a regime with an established process can take three to six months from submission to approval. A full CASP authorisation under MiCA, or a major payment-institution licence in Singapore, typically takes six to twelve months, and can extend beyond a year if the regulator requests supplemental information or if the business must remediate deficiencies in its compliance, capital or insurance arrangements. Early preparation – including securing indicative insurance quotes, completing technical and financial audits, and engaging local counsel where required – shortens the timeline materially.

Which jurisdiction suits my crypto business?

The answer turns on where your clients are, where you can bank, and where your operational and tax structure sits most efficiently. For a business targeting institutional clients in the EU, a MiCA CASP licence in a member state with robust banking infrastructure (for example, Lithuania or Malta transitioning to the MiCA regime) is often the most credible route. For a business focused on the Middle East and Asia, VARA in Dubai or the ADGM framework in Abu Dhabi provides a clear regulatory path and access to regional banking. Businesses serving a global retail base often need licences in multiple jurisdictions. The licensing strategy should be developed in parallel with the banking and tax strategy, because a licence is only valuable if you can open accounts and route funds efficiently.

Do I need a separate custody licence?

In many regimes, custody is one activity within a broader VASP or payment-services licence, so there is no separate "custody licence" as such – the business applies for the general licence and specifies custody as one of the regulated activities it will perform. Under MiCA, custody and administration of crypto-assets is a distinct regulated activity, and a CASP authorised for that activity may provide custody. In some jurisdictions (for example, certain US states, or Switzerland), a business providing custody may need a trust-company or banking licence depending on the structure. The question is not always "do I need a custody licence" but "does my operating model constitute regulated custody, and if so, which licence or authorisation covers it".

Related at OBOLUS

About OBOLUS. OBOLUS is an independent digital-asset law boutique acting only for businesses. We advise exchanges, custodians, token issuers and funds on licensing across 70+ jurisdictions, on disputes and on-chain asset recovery across 25+ forums, and on the tax, banking and compliance that sit around them. Digital assets are the entirety of our practice, and we work only with operators who understand the technology and need the legal architecture to match. We separate the regulated custody perimeter from the technical custody model before you commit capital or submit an application, and we coordinate the insurance and capital structuring that underpins authorisation in every major hub. To map your licensing path or to review your custody-insurance package before submission, contact info@oboluslaw.com.

By Aisha Tan, Licensing & Jurisdictions Analyst – Aisha specialises in the regulatory authorisation process for digital-asset service providers, with a focus on custody, capital adequacy and the insurance requirements that underpin VASP and CASP licences across EU, MENA and Asia-Pacific jurisdictions.

This publication is general information about the law and does not constitute legal advice. It is not a substitute for advice tailored to your circumstances. OBOLUS accepts no liability for action taken or not taken on the basis of this material. For advice on your situation, contact info@oboluslaw.com.

Discuss your matter with OBOLUS

Tell us about the business, the jurisdictions in scope and the deadline. Every brief is confidential and we reply within two hours.

Or email info@oboluslaw.com. We treat your brief as confidential.

Tell us the task — we'll map your options in 30 minutes.

Fixed-fee packages with defined scope and SLAs. The first call is free and under NDA. Business clients only.

Map your optionsinfo@oboluslaw.com · reply < 2 hours