Lesson 9 of 14 · about 55 minutes · Updated November 15, 2026
Cyber Insurance
You'll be able to
- Explain what first-party and third-party cyber coverages pay for, and the common exclusions, sublimits, retentions and waiting periods that limit them.
- Describe the U.S. cyber insurance market's size, loss ratio and price cycle from 2021 to 2026 using NAIC and Marsh data.
- Prepare and verify an insurance application so that control answers are backed by evidence, and explain the risk shown by Travelers v. International Control Services.
- Compare old war exclusions with the Lloyd's state-backed cyber attack requirements and explain what the Merck and Mondelez cases mean for buyers.
- Plan the claims process, ransom payment authority and sanctions screening before an incident, and size limits from the organization's own loss scenarios.
- Weigh the evidence on systemic cyber risk, a federal backstop, public entity pools and whether insurance improves security.
Start here: the short version
- Cyber insurance pays for some of the money side of an incident, not the incident. A typical policy covers first-party costs (response experts, forensics, notification, data restoration, lost income, extortion) and third-party claims (lawsuits, regulatory defense and, where the law allows, fines and card brand assessments). It does not restore patient care, rebuild trust or satisfy a regulator's duty to protect data.1
- The U.S. market is large and, for buyers, currently soft. The NAIC counted about $9.46 billion of U.S. cyber direct written premium in 2025, up about 3.5 percent from 2024, and an industry loss ratio of 53. Marsh reported U.S. cyber rates flat in the third quarter of 2026 after a long run of cuts, following the steep increases of 2021 and 2022.23
- The application is a sworn description of your controls. In 2022 Travelers and International Control Services agreed to have a court declare a cyber policy void from the start after Travelers alleged the company had misstated its use of multifactor sign-in. Every answer must be checked against evidence.4
- War and state-backed attack exclusions are the biggest wording fight. A New Jersey appeals court held on May 1, 2023 that an old "hostile or warlike action" exclusion did not bar Merck's NotPetya property claim; Lloyd's has required its syndicates to carry explicit state-backed cyber attack exclusions on standalone cyber policies since March 31, 2023.56
- Systemic events test the model. Modelers put insured losses from the July 19, 2024 CrowdStrike outage at about $400 million to $1.5 billion, a fraction of the economic loss. Treasury has studied a federal backstop for catastrophic cyber events since 2022 and, as of 2026, has not proposed one.78
- Insurance is a financing tool that can push better security, weakly. Research such as the 2023 RUSI study found no strong evidence that insurance drives ransom payments, and found insurers' influence on security real but underused. Treat cover as one control among many, bought with the board's eyes open.9
Rules, wordings and data described: NAIC 2026 Report on the Cybersecurity Insurance Market (2025 data) and the 2025 report (2024 data); Marsh Global Insurance Market Index for the third quarter of 2026 (released Oct. 21, 2026); Lloyd's Market Bulletin Y5381 (August 2022, requirements effective March 31, 2023) and the Lloyd's Market Association clauses LMA5564 to LMA5567 and their A and B versions; Merck & Co. v. ACE American Insurance Co., N.J. Super. Ct. App. Div., Nos. A-1879-21 and A-1882-21 (May 1, 2023); Treasury OFAC ransomware advisory of Sept. 21, 2021; New York DFS Insurance Circular Letter No. 2 (2021); SEC Release No. 33-11216 (2023); GAO-22-104256 (June 21, 2022) and GAO-25-108748 (Sept. 23, 2025); Idaho Code Title 74, Chapter 1, as amended through the 2026 session. All read Nov. 15, 2026. Policy wordings differ by insurer and change at every renewal, market data are revised, and rules change; read your own policy and the regulator's current text before relying on any point.
Before you change anything. Plan changes to production systems, test them first, keep a way back, and follow your change management process. For clinical, plant-floor or other safety systems, involve the vendor, your safety lead and qualified engineers (see modules 109 and 111).
It is the third day of a ransomware incident at a regional health system. The chief financial officer has the cyber policy open on one screen and the claims adjuster's email on the other. The policy has a $25 million limit, but the line for "cyber extortion" says $5 million, the business interruption section starts paying only after a 12-hour waiting period, and the response firm the security team called on the first night is not on the insurer's approved panel. The insurer's letter asks for a copy of the renewal application signed eight months ago, which said multifactor sign-in protected "all remote access." The attackers came in through a vendor's remote support account that did not have it.
None of those facts is unusual. Each of them is decided long before the incident: when the organization chose its limits, read (or did not read) the exclusions, filled in the application, picked its response firms and decided who could approve a ransom. This lesson is about those decisions. It is written for the people who make or oversee them in a large or regulated organization: chief financial officers and risk managers who buy insurance, CISOs who answer underwriters' questions, general counsel's offices and compliance officers who read the wording, audit and risk committee members who approve the program, public officials who run risk pools, and the reporters, legislators and advocates who ask why a public body was or was not covered.
Insurance basics for smaller organizations are in Continuity, Recovery and Cyber Insurance in module 73, and this lesson does not repeat them. Risk transfer as one of the four risk responses is explained in Risk Management in Plain Terms, and the incident response machinery a policy plugs into is in Detection and Incident Response. Inside this module, the controls underwriters ask about are measured in Metrics and Reporting to Leadership, the money for premiums and retentions is planned in Budgets, Staffing and the Business Case, insurance in deals is in Security in Mergers and Acquisitions, and the recovery plans that make a business interruption claim smaller are in Operational Resilience and Crisis Management.
What a cyber policy is
A cyber insurance policy is a contract in which an insurer agrees to pay defined costs and claims that follow defined events, such as a "security failure," a "privacy event," a "cyber extortion threat" or a "system failure," in exchange for a premium, subject to limits, retentions, conditions and exclusions. There is no standard national form the way there is for homeowners or commercial general liability insurance. Each insurer writes its own wording, and large buyers often negotiate manuscript changes. Two policies with the same limit can pay very different amounts after the same incident.
The building blocks
- Insuring agreements. Each coverage part (incident response, business interruption, extortion, privacy liability and so on) has its own promise to pay. Read them one by one; a coverage that is not in an insuring agreement is not in the policy.
- Limit of liability. The most the policy pays, usually an annual aggregate across all claims, with smaller sublimits inside it for particular coverages (often ransomware, dependent business interruption, social engineering fraud, regulatory fines or card brand assessments).
- Retention. The insured's share before the insurer pays, the cyber equivalent of a deductible. Large organizations often carry retentions of $1 million to $10 million or more; business interruption usually has a separate waiting period measured in hours.
- Claims-made trigger. Liability parts usually cover claims first made during the policy period and reported within it or a short window after it, about events after a retroactive date. Switching insurers without matching the retroactive date can leave a gap for old intrusions found late.
- Conditions. Duties the insured must meet: prompt notice, cooperation, using approved vendors or getting the insurer's consent before spending money, not admitting liability, preserving evidence, and sometimes notifying law enforcement.
- Exclusions. What is not covered: war and state-backed attacks, failures of power, telecommunications or other infrastructure, prior known incidents, bodily injury and property damage (often), contractual liability beyond what the law would impose, the cost of improving systems ("betterment"), and fines that the law will not allow to be insured.
- The application. The questionnaire, attachments, scan results and any warranty letters. Most policies make the application part of the contract.
Standalone policies, endorsements and towers
Cyber coverage reaches buyers in three forms. A standalone cyber policy is a dedicated contract. An endorsement or package adds a cyber section to another policy, such as a business owner's or property policy; these are common for small businesses and usually carry low limits. Large organizations buy a tower: a primary policy for the first layer of limit and one or more excess policies stacked above it, each paying only after the layers below are exhausted. The NAIC's 2026 report, covering 2025, found that primary policies were about 64 percent of U.S. cyber premium, excess about 32 percent and endorsements about 4 percent, even though endorsements were roughly half of all policies in force.210 In other words, most policies are small add-ons; most premium dollars buy dedicated cover for larger organizations.
Excess layers usually "follow form," meaning they adopt the primary policy's terms, but not always; an excess insurer may have its own war exclusion, its own sublimits or its own notice rules. A risk manager who reads only the primary wording has not read the program.
What cyber insurance covers
The Federal Trade Commission's plain-language guide for businesses divides cyber insurance into first-party coverage, which pays the insured's own losses, and third-party coverage, which pays when others make claims against the insured. Its first-party list includes legal help to determine notification and regulatory obligations, recovery and replacement of lost or stolen data, customer notification and call center services, lost income from business interruption, crisis management and public relations, cyber extortion and fraud, forensic services, and "fees, fines, and penalties related to the cyber incident." Its third-party list includes payments to affected consumers, claim and settlement expenses, defamation and copyright or trademark infringement, litigation and regulatory inquiry costs, settlements, damages and judgments, and accounting costs.1 Real policies use their own names, but the map is close to this.
First-party coverages
| Coverage | What it pays for | What to check in a large program |
|---|---|---|
| Incident response (often called breach response or event management) | The law firm that coordinates the response and preserves privilege, digital forensics and incident response (DFIR) firms, crisis communications. | Whether you must use panel firms, whether your own retained firms can be pre-approved, hourly rate caps, and whether these costs erode the limit or sit outside it. |
| Notification and remediation | Letters, call centers, credit monitoring and identity restoration for affected people. | Whether it is counted by dollars or by number of people notified; whether voluntary notices (not legally required) are covered. |
| Data and system restoration | Recreating or restoring data and software from backups or other sources. | Hardware replacement ("bricking") is often excluded or sublimited; betterment (making systems better than before) usually is not covered. |
| Business interruption (BI) | Lost net income and extra expense while the insured's own systems are down because of a covered event. | Waiting period, period of restoration, how loss is calculated, and whether "system failure" (non-malicious outages) is covered as well as security failures. |
| Dependent or contingent business interruption | The same, when a vendor's or cloud provider's outage stops the insured's business. | Often sublimited; may cover only named providers or only security failures, not vendor system errors. Critical after the 2024 CrowdStrike and Change Healthcare events. |
| Cyber extortion | Experts to assess and negotiate a threat and, where lawful, the payment itself. | Sublimits and coinsurance for ransomware, insurer consent before payment, sanctions screening. |
| Social engineering and funds transfer fraud | Money lost when staff are tricked into sending payments, or when attackers move money directly. | Frequently placed on a crime policy instead; low sublimits; call-back verification conditions. |
| Reputational harm | Some wordings pay lost income from customers leaving after a publicized incident. | Narrow, time-limited, hard to prove. |
Third-party coverages
| Coverage | What it pays for | What to check |
|---|---|---|
| Network security and privacy liability | Defense and damages when customers, patients, business partners or class action plaintiffs sue over a breach or the spread of malware. | Whether defense costs erode the limit; who controls the defense; consent to settle (the "hammer clause"). |
| Regulatory defense and penalties | Responding to investigations by the FTC, HHS Office for Civil Rights, state attorneys general, the SEC or foreign data protection authorities, and fines or penalties "where insurable by law." | Insurability of fines varies by state and by country and by type of penalty; the policy usually applies the law most favorable to coverage, but some penalties cannot be insured anywhere. |
| Payment card industry (PCI) fines and assessments | Contractual assessments and fines the card brands pass through the acquiring bank after a card data breach, and related forensic costs. | Usually sublimited; may require that the insured was PCI DSS compliant at the time (see Certifications and Government Authorizations). |
| Media liability | Defamation, infringement and similar claims from the organization's online content. | Overlap with other media or professional policies. |
What it usually does not cover
The common gaps are as important as the coverages. Most cyber policies exclude or limit:
- Bodily injury and property damage. A hospital whose ransomware event delays care, or a plant whose control systems damage equipment, may find those losses left to general liability and property policies, which have their own cyber exclusions. Manufacturers should read Threats and Real Cases in module 111 with their property wording beside it.
- War and state-backed attacks, covered in detail below.
- Infrastructure failure: outages of power, internet backbone, satellite or telecommunications services not under the insured's control.
- Prior known acts: incidents the organization knew about before the policy began.
- Betterment: the cost of upgrading security after an incident, even when the insurer's forensic firm recommends it.
- Lost value of intellectual property and future lost profits beyond the restoration period.
- Uninsurable fines, and some criminal penalties in any case.
- Contractual liability the insured took on by contract beyond what the law would impose anyway, with exceptions for PCI and sometimes for confidentiality promises.
NIST's enterprise risk guidance states the general point plainly: transferring risk shares "a portion of the consequences with another party," and "there are often consequences that cannot be transferred, like the loss of customer trust."11
The market: from hard to soft
Cyber insurance prices swing more than most commercial lines because the risk is new, losses are lumpy and insurers' data are thin. Buyers who understand the cycle can time their asks: better terms and higher limits when the market is soft, tighter controls and higher retentions when it is hard.
How big the U.S. market is
The National Association of Insurance Commissioners (NAIC), the organization of state insurance regulators, publishes an annual report built from the cyber supplement that insurers file with their statutory financial statements. Its 2026 report, covering calendar year 2025, found:2
- About $9.46 billion of U.S. cyber direct written premium in 2025 including surplus lines insurers domiciled outside the United States (alien surplus lines), up about 3.5 percent from about $9.14 billion in 2024. U.S.-domiciled insurers wrote about $7.46 billion, up from $7.08 billion.
- About 5.84 million policies in force at domestic insurers, up 33 percent, with endorsement policies up 53.8 percent and primary policies up 13.6 percent.
- An industry loss ratio, counting losses and defense and cost containment expenses, of 53 in 2025, up from 48.7 in 2024. A loss ratio of 53 means that for every $100 of premium earned, about $53 went to claims and claim defense, before brokers' commissions, the insurer's own expenses and profit.
- 56,323 reported claims at domestic insurers, up 20 percent, and more than six times the 2017 level.
- The three largest groups by 2025 cyber premium were Chubb (about $629 million), Beazley (about $571 million) and Travelers (about $544 million); the top 20 groups wrote about 77 percent of the market.
The year before, the NAIC's 2025 report had recorded the market's "first ever reduction" in written premium: about $9.14 billion in 2024, down about 7 percent from about $9.84 billion in 2023, while claims rose by almost 40 percent to nearly 50,000.10 Trade press coverage of the 2026 report stressed that domestic premium has nearly tripled since 2020 ($2.75 billion) and that surplus lines insurers, which are less tightly regulated on rates and forms, write nearly two-thirds of U.S. cyber premium.12 These are U.S. figures; the global market is larger, and many large U.S. buyers place part of their towers in London.
The hard market of 2021 and 2022
Ransomware changed the market in 2020 and 2021. As claims for extortion and business interruption mounted, insurers raised prices, cut limits, added ransomware sublimits and coinsurance, and began refusing to quote organizations that could not show basic controls. Marsh, the largest broker, reported in its Global Insurance Market Index that U.S. cyber prices rose 130 percent in the fourth quarter of 2021, up from 96 percent in the third quarter, and U.K. prices rose 92 percent, which Marsh tied to "the ongoing increase in the frequency and severity of ransomware claims."13 Those were averages; some buyers saw renewals double or triple while their limits were cut in half. This was the period in which multifactor sign-in, endpoint detection and response, and offline backups moved from recommendations to conditions of cover.
The soft market of 2024 to 2026
Higher prices and tighter underwriting brought profits and new capital, and prices began to fall. Marsh's index for the third quarter of 2026, released Oct. 21, 2026, reported global cyber rates down 4 percent, "the thirteenth consecutive quarter of declines," and U.S. cyber rates flat "after an extended period of declines."3 The NAIC's 2026 report cited the first quarter of 2026 as the "eighth consecutive quarter of U.S. cyber rate cuts."2 At the same time, the NAIC data show the loss ratio rising (48.7 to 53) and average severity of large-business claims about doubling, from about $2.2 million in 2024 to about $4.4 million in 2025.2 Falling prices with rising losses is the classic setup for the next hardening. A risk manager who locks in broad wording, higher limits and multi-year terms in a soft market is buying protection against that turn.
What the claims look like
Claims data come mostly from insurers and brokers, and each describes its own book, not the whole market. With that caveat, they show where the money goes:
- Coalition, an insurer that writes mostly small and mid-sized businesses, reported in its 2026 Cyber Claims Report (2025 data, released March 5, 2026) that business email compromise and funds transfer fraud together were 58 percent of its claims; ransomware was the costliest claim type, at an average loss of about $269,000; initial ransom demands rose 47 percent; and a record 86 percent of businesses that received a demand refused to pay.14
- The FBI's Internet Crime Complaint Center recorded business email compromise losses of about $2.77 billion reported to it in 2024, and $16.6 billion in total reported internet crime losses.15 Much of that loss falls on crime or cyber policies, or on no policy at all.
- For large companies, a single event can consume a tower. Merck reported that the 2017 NotPetya attack cost it about $260 million in lost sales in 2017 and $285 million in expenses net of insurance recoveries of about $45 million, and its property insurers fought the rest of the claim for six years (below).16 MGM Resorts said in an October 2023 filing that a cyberattack would cut its quarterly property earnings by about $100 million and that it expected its cyber insurance to be sufficient to cover the financial impact.17

What underwriters ask
Underwriting is how an insurer decides whether to offer cover, at what price and on what terms. For a large organization it usually involves a long application, supplemental questionnaires on ransomware and specific technologies, an external scan of the organization's internet-facing systems, and an underwriting meeting at which the CISO presents the program. The questions have converged since 2021 on a short list of controls that claims data tie to losses.
The controls underwriters care about
| Control | What underwriters typically ask | Where this course explains it |
|---|---|---|
| Multifactor sign-in (MFA) | Is it enforced for all remote access, including vendors; for email; for all privileged and administrator accounts; for cloud consoles and backups? Is it phishing-resistant? | 108-05 Identity and Access |
| Endpoint detection and response (EDR) | What share of workstations and servers run it; is it monitored around the clock, by staff or a managed service? | 108-10 |
| Backups | Are backups offline or immutable, separated from the main identity system, encrypted, and tested by actual restores? How long to restore critical systems? | 108-07 |
| Privileged access management | How many domain administrator accounts; are service accounts controlled; is there a vault for privileged credentials? | 108-05 |
| Patching and end-of-life systems | How fast are critical and known-exploited vulnerabilities patched; are internet-facing devices on CISA's Known Exploited Vulnerabilities catalog fixed within its due dates; what unsupported systems remain and how are they isolated?18 | 108-06 |
| Email security | Filtering, link and attachment sandboxing, DMARC, banners on external mail, payment verification procedures for wire changes. | 110-06 |
| Remote desktop and remote access | Is remote desktop protocol exposed to the internet anywhere; how is vendor remote access controlled? | 108-06, 111-04 |
| Incident response plan and exercises | Written plan, tested in the last year, with the insurer's hotline in it. | 108-10, 112-12 |
| Training and phishing tests | Frequency and results. | 108-09 |
| Vendor and cloud dependencies | Which critical vendors, how they are assessed, what contracts require. | 108-08, 112-05 |
The focus on remote access with multifactor sign-in is not arbitrary. UnitedHealth Group's chief executive told the Senate Finance Committee in May 2024 that the criminals behind the Change Healthcare attack used compromised credentials to get into a remote access portal that did not have multifactor authentication.19 That single control gap is exactly what most applications ask about first.
Do the controls predict losses?
Some evidence says yes, with limits. In April 2023 Marsh McLennan published research that combined its claims data with answers to its cybersecurity self-assessment questionnaire and examined twelve control categories commonly required by insurers, ranking those most strongly associated with fewer cyber events. It found that automated hardening techniques had "by a wide margin" the strongest association, that organizations patching high-severity vulnerabilities within seven days roughly halved their probability of an event (only 24 percent of the organizations studied did so), and that multifactor sign-in reduced risk only when deployed broadly, across critical data, all remote access and administrator accounts.20 This is a broker's study of its own data, not peer-reviewed research, and association is not proof of cause. It is still more evidence than most security spending decisions get, and it explains why underwriters ask about breadth of coverage ("what percentage of...") rather than whether a tool has been bought.
The application is a legal statement
Most cyber policies say the insurer relied on the application, and that the application is part of the policy. If an answer was materially false, the insurer may try to rescind the policy (treat it as if it never existed) or deny a claim. State law decides what counts as a material misrepresentation and whether the insurer must show it relied on the answer or that the answer was made with intent to deceive, and the rules vary.
The leading example is Travelers Property Casualty Co. of America v. International Control Services, Inc., No. 22-cv-2145, in the U.S. District Court for the Central District of Illinois. International Control Services, an electronics manufacturer in Decatur, Illinois, bought a Travelers cyber policy running from April 4, 2022 to April 4, 2023. Its president signed the application and a multifactor authentication attestation in March 2022. After a ransomware event in May, Travelers alleged that the company had "misrepresented its use of MFA on the insurance application" and that the misinformation "materially affected the acceptance of the risk." Travelers asked the court in July 2022 to rescind the policy. On Aug. 26, 2022 the parties filed a joint stipulation asking the court to declare the policy "null and void, from its inception," with no coverage for any past, present or future claim, and to dismiss the case with prejudice, each side paying its own costs.4 There was no court finding that the company lied; the company gave up the policy rather than fight. The lesson is the same either way: the person who signs the application is making statements about controls they may not personally run, and the organization can lose the whole policy over one wrong box.
Warranties, knowledge and severability
Three pieces of wording decide how badly a wrong answer can hurt:
- Warranty statements. Some insurers ask a senior officer to sign a letter warranting that no one in a named group knows of any incident or circumstance likely to give rise to a claim. A claim arising from something that was known can be excluded.
- Knowledge group. Whose knowledge counts: only the chief executive, CFO, general counsel's office and CISO, or any employee? A narrow group protects the organization when a junior employee knew of a problem.
- Severability and non-rescission. Severability means one person's false statement is not imputed to other insureds. Some large buyers negotiate a non-rescission clause, under which the insurer can deny a specific claim affected by a misstatement but cannot void the whole policy. In a soft market, ask for it.
Controlling the application
A large organization should treat its insurance application like a public filing: drafted by the people who know, reviewed by people who check, and approved by an officer who knows what was checked. A workable process:
- The risk manager or broker sends the full application and supplementals to the CISO's office at least 90 days before renewal.
- Each control question is assigned to a named control owner, who answers it and attaches evidence: a coverage report from the identity system, an EDR console count, a restore test record, a vulnerability aging report.
- Answers that are true only in part are written as true only in part, with the exceptions listed (for example, "MFA enforced on all remote access except two legacy OT vendor connections, which are isolated and scheduled for replacement by [date]"). An honest exception costs less than a rescinded policy.
- Internal audit or the second line (risk or compliance) samples the evidence, as it would for a regulatory certification. Lesson Internal Audit, SOX and the Three Lines explains how.
- The signing officer receives a short memo listing what was verified, what was qualified and who verified it.
- The final application, attachments and evidence are kept with the policy for at least the life of the policy plus the period in which claims could be made.
- Material changes during the policy year (a control removed, an acquisition, a new incident) are reviewed against the policy's notice requirements.
War and state-backed attack exclusions
Insurers have excluded war losses for more than a century, because a war can damage everything at once and no private pool can pay for that. The question since 2017 has been how an old war exclusion applies to a cyberattack that a government carried out, or is said to have carried out, without any declared war, and that damaged companies far from the target.
NotPetya and two coverage fights
On June 27, 2017 destructive malware disguised as ransomware spread from a compromised Ukrainian tax software update to companies around the world. The White House attributed it to the Russian military in February 2018. It could not be undone by paying. Two large U.S. companies then sued their insurers, and both cases were about property policies with old war exclusions, not modern cyber policies.
Mondelez International v. Zurich American Insurance. Mondelez, the snack maker, sought about $100 million under a property policy (it did not have a separate cyber policy), and Zurich invoked the policy's exclusion for "hostile or warlike action" by a "government or sovereign power." The case was litigated in Illinois state court for about four years and settled in late October 2022 on undisclosed terms, before any final ruling on the exclusion.21
Merck & Co. v. ACE American Insurance Co. Merck's property program had $1.75 billion in limits above a $150 million deductible, provided by more than 30 insurers; by the time of the appeal, the insurers still contesting the claim faced about $700 million, a little under 40 percent of the program.22 A New Jersey trial court ruled in January 2022 that the war exclusion did not apply. On May 1, 2023 the Appellate Division affirmed in Nos. A-1879-21 and A-1882-21. Writing for the court, Judge Currier held that the exclusion for "hostile or warlike action in time of peace or war" by a government or sovereign power "requires the involvement of military action," and that its terms were meant to cover "actions clearly connected to war or, at least, to a military action or objective." The court noted that the insurers had not updated their language despite knowing that cyberattacks were a growing risk, and that a suspected Russian attack on a Ukrainian target was not enough to bring an attack on a pharmaceutical company within the exclusion.5 The New Jersey Supreme Court agreed in July 2023 to hear the insurers' appeal, but Merck and the remaining insurers settled in early January 2024 on confidential terms before the court ruled.22
The two cases settled the practical question even without a final high-court ruling: old war exclusions written for armies are a weak basis for denying cyber claims, and insurers responded by rewriting the exclusions.
Lloyd's and the new state-backed attack exclusions
Lloyd's of London, a marketplace where many large U.S. cyber towers are partly placed, issued Market Bulletin Y5381 in August 2022. It requires every Lloyd's syndicate writing standalone cyber attack policies to include a clause excluding liability for losses from state-backed cyber attacks, at the inception or renewal of each policy from March 31, 2023.6 The bulletin warned that such risks could "greatly exceed what the insurance market is able to absorb."23 At a minimum, the exclusion must:
- Exclude losses arising from a war, declared or not, where the policy has no separate war exclusion.
- Exclude, subject to carve-backs the syndicate may agree, losses from state-backed cyber attacks that "significantly impair the ability of a state to function" or "significantly impair the security capabilities of a state."
- Be clear about whether it excludes computer systems located outside the affected state.
- Set out a clear basis on which the parties will agree whether an attack was state-backed (attribution).
- Define all key terms clearly.6
The Lloyd's Market Association had already published four model clauses, LMA5564 to LMA5567, which range from a full exclusion of cyber operations by states to versions that keep more cover. After the bulletin it issued revised "A" versions that meet the bulletin's requirements and "B" versions that leave out the attribution clause, so a syndicate using a B clause must supply its own attribution method.24 U.S. insurers outside Lloyd's have written their own versions, with differing thresholds: some exclude only attacks with a "major detrimental impact" on a state, some turn on a widespread event across many organizations.
What a buyer should negotiate
- Who decides attribution, and on what evidence. Prefer wording that relies on a formal attribution by the government of the state where the insured's affected systems are located, with the burden of proving the exclusion on the insurer.
- Collateral damage. Ask whether the exclusion applies when the insured is not the target, only a bystander hit by spread from an attack on another country.
- Essential service carve-backs. Many clauses exclude attacks that impair a state's "essential services." For a hospital, utility or bank, that definition may describe the insured itself.
- Consistency across the tower. If the primary and each excess layer use different clauses, one event can be covered in one layer and excluded in another.
- Property and other policies. NotPetya losses were fought under property policies. Check the cyber exclusions in the property, general liability, crime and kidnap and ransom policies at the same time, so a loss does not fall between them. New York's Department of Financial Services has told the insurers it regulates to eliminate "silent" cyber risk (cyber losses covered by policies that do not mention cyber) by making clear whether each policy covers or excludes cyber losses.25
Ransomware terms and paying ransoms
Ransomware drove the 2021 to 2022 repricing and still drives the largest claims. Policies now treat it separately in several ways:
- Ransomware sublimits. A cap on all loss arising from a ransomware event (extortion, business interruption, restoration and sometimes liability) below the full policy limit.
- Coinsurance. The insured pays a percentage of each ransomware loss above the retention, for example 25 or 50 percent, so it keeps a stake in the outcome.
- Conditions tied to the controls on the application, such as offline backups or MFA, sometimes written as a condition that reduces or removes cover if the control was not in place.
- Consent: no payment without the insurer's prior written consent.
In the soft market of 2024 to 2026 many large buyers with strong controls have been able to remove or raise ransomware sublimits. Ask; the answer depends on the evidence you can show.
Sanctions and the law on paying
Paying a ransom is not generally illegal in the United States, but paying a sanctioned person is. The Treasury Department's Office of Foreign Assets Control (OFAC) issued an updated advisory on Sept. 21, 2021 that says OFAC "may impose civil penalties for sanctions violations based on strict liability," meaning without proof the payer knew the recipient was sanctioned, and that the U.S. government "strongly discourages all private companies and citizens from paying ransom or extortion demands." It names cyber insurance firms, financial institutions and incident response firms among those that may facilitate payments and so may risk violating OFAC rules. It treats a prompt, complete report to law enforcement and full cooperation as "a significant mitigating factor" in any enforcement decision.26
New York's Department of Financial Services went further for insurers it regulates. Its Insurance Circular Letter No. 2 (2021), issued Feb. 4, 2021, set out a seven-part Cyber Insurance Risk Framework: a formal cyber insurance risk strategy, elimination of silent cyber risk, evaluation of systemic risk, rigorous measurement of each insured's risk, education of insureds and brokers, in-house cybersecurity expertise, and a requirement in policies "that victims notify law enforcement." It states that "DFS recommends against making ransom payments."25 Some states have barred state agencies and local governments from paying ransoms; check your own state's law, since Idaho had no such statute that this lesson could find as of Nov. 15, 2026.
CISA, the FBI and their partners also advise against paying and publish a free response checklist.27 Coalition's 2026 claims data, in which 86 percent of its policyholders that received demands refused to pay, suggest that good backups and response plans make refusal practical for most organizations.14
Who decides whether to pay
The policy does not decide whether to pay; the organization does, with the insurer's consent needed to be reimbursed. A large organization should decide in advance, in writing:
- Who has authority to approve a ransom payment (usually the chief executive with the board chair or a committee, never the CISO alone).
- What facts the decision requires: whether backups can restore operations and how fast, whether life or safety is at risk, whether data theft makes payment pointless, the sanctions screening result and law enforcement's input.
- How the decision will be recorded and disclosed. A public company that pays may need to consider the payment in its materiality decision under Form 8-K Item 1.05, covered in 108-04.28

Panels, consent and the claims process
Large cyber claims are managed claims. The insurer, the insured, the coordinating law firm, forensic investigators, restoration firms, negotiators, forensic accountants and sometimes regulators are all involved for months. How it goes depends on decisions made before the incident.
Panel vendors and consent
Most policies offer a panel of pre-approved vendors: law firms that coordinate breach response, forensic firms, notification and call center vendors, public relations firms and ransomware negotiators. Using panel firms usually means pre-agreed rates and no argument over whether costs are covered. Using other firms usually requires the insurer's prior written consent, and the insurer may pay only the panel rate.
Large organizations often have their own retained incident response firm and law firm, already familiar with their systems. The fix is to ask the insurer to add those firms to the policy as approved vendors at renewal, in writing, with their rates. Do it before an incident; the first night of a ransomware event is the wrong time to negotiate. Put the insurer's hotline number and the approved vendor list into the incident response plan described in 108-10.
Researchers who studied the role of insurers in incident response describe this as both the strongest part of the product and a source of conflict: insurers fund expert response that many organizations could not organize alone, but when the insurer picks the firm, the firm's incentives may lean toward the insurer's interest in a cheaper outcome.29 Know whom each vendor works for, and keep the organization's own legal and security leadership in charge of decisions.
A claim, step by step
- Hour 0 to 24: notice. Call the insurer's incident hotline and give written notice as the policy requires ("as soon as practicable" or within a set number of days). Late notice is one of the most common reasons for disputes. Notify every layer of the tower, not just the primary.
- Day 1: engage approved vendors. The coordinating law firm engages forensic and other firms so that their reports are prepared at its direction. Confirm in writing which firms the insurer has approved.
- Day 1 to 3: preserve and report. Preserve logs, images and the ransom note. Report to the FBI (field office or ic3.gov) and to CISA; NYDFS-regulated insurers' policies may require law enforcement notice, and OFAC treats it as mitigating.2625
- Days 1 to 30: regulatory clocks. Track the deadlines that the incident starts: Form 8-K Item 1.05 four business days after a materiality determination, NYDFS 72 hours, bank regulators 36 hours, state breach notice laws (covered in 108-11). Tell the insurer before any public statement, to avoid an argument that the statement admitted liability.
- Weeks 1 to 8: restore and document. Record every cost, every hour of downtime and every manual workaround. Business interruption begins after the waiting period and is measured against what income would have been.
- Months 2 to 12: proof of loss. Submit a sworn proof of loss with forensic accountants' calculations. The insurer's accountants will test it. Large business interruption claims often take a year or more.
- Months 6 to 24: liability claims and settlement. Class actions and regulatory investigations follow, sometimes for years, and draw on the same limit unless defense sits outside it.
Business interruption is where the money is
For a large organization, lost income usually dwarfs notification costs. Three things make business interruption claims smaller and easier to prove: knowing in advance what each hour of downtime costs for each critical process (the business impact analysis in Operational Resilience and Crisis Management), keeping manual workarounds that keep revenue flowing, and recording, from the first hour, which systems were down, which revenue was lost and which extra costs were spent to keep going. Agree in advance with the insurer, if possible, on the method for measuring loss, and engage forensic accountants early.
When insurer and insured disagree
Disputes over large claims usually concern the size of the business interruption loss, whether costs were reasonable or approved, whether an exclusion applies, and whether the application was accurate. Most are settled by negotiation, appraisal or mediation, as Mondelez and Merck eventually were. Policies may require arbitration, set the governing law, or require suit within a limited period. The risk manager should know these terms before the claim, not after.
Systemic risk and the limits of private insurance
Most insurance works because losses are independent: one house burns, not all of them. Cyber risk is different. A flaw in widely used software, an outage at a dominant cloud or security vendor, or a worm can hit thousands of insureds at once. Insurers and their regulators call this systemic or catastrophic cyber risk, and it shapes what the market will sell.
CrowdStrike, July 19, 2024
On July 19, 2024 a faulty content update to CrowdStrike's Falcon endpoint security software crashed Windows computers around the world. Microsoft estimated that it affected 8.5 million Windows devices, "less than one percent of all Windows machines."30 Airlines, hospitals, banks and broadcasters lost systems for hours to days. It was not an attack, so the question for insurance was whether policies covered "system failure" and dependent business interruption from a vendor's non-malicious error. Early estimates of insured losses were modest compared with the disruption:
- CyberCube, a cyber risk modeling firm, estimated global insured losses of $400 million to $1.5 billion.7
- Guy Carpenter, a reinsurance broker, estimated $300 million to $1 billion, "a sizable but manageable insured loss."31
- Parametrix, a parametric cloud outage insurer, estimated about $5.4 billion of direct losses to Fortune 500 companies alone, excluding Microsoft, of which it estimated only $540 million to $1.08 billion would be insured.732
These were model estimates made in the weeks after the event, not counts of paid claims, and final figures have not been published in a single public source that this lesson could find. The gap between economic and insured loss is the point: most of the cost fell on organizations whose policies did not cover non-malicious vendor outages, whose waiting periods outlasted the outage, or who had no cyber cover at all.
Jaguar Land Rover, 2025
In late August 2025 a cyberattack forced Jaguar Land Rover, Britain's largest carmaker, to shut down its IT systems and halt production at its U.K. plants; phased production restarted on Oct. 8, 2025, more than five weeks later. The U.K. government guaranteed a £1.5 billion loan so the company could keep paying suppliers.33 The Cyber Monitoring Centre, an independent U.K. body that rates cyber events, estimated the total financial damage across the U.K. at about £1.9 billion (a modeled range of £1.6 billion to £2.1 billion), affecting more than 5,000 organizations, and rated it a category 3 systemic event.34 Insurance Business, a trade publication, reported that the company had no cyber policy in place when it was attacked because a placement through its broker had not been completed; the company has not confirmed this publicly in any source this lesson could find, so treat it as a press report.35 The case is a U.K. one, but it shows what U.S. boards should ask: if our largest supplier or our own plants stopped for five weeks, what would insurance pay, and what would we do for cash?
Is there a federal backstop?
The federal Terrorism Risk Insurance Program, created in 2002, shares terrorism losses between insurers and the government. Treasury clarified in 2021 that it can cover terrorism losses under eligible cyber policies, but GAO has explained that a cyberattack may not meet the program's tests: it may not be "violent" or dangerous to human life, property or infrastructure, may not be aimed at coercing the U.S. population or government, and may not cause damage in the right places.36 A state-backed attack is not terrorism under the program at all.
Federal work on a separate cyber backstop has moved slowly:
- June 21, 2022. GAO's report GAO-22-104256 found that insurers were excluding cyber warfare and infrastructure outages to limit losses from systemic events, that the terrorism program might not cover catastrophic cyberattacks, and that neither Treasury's Federal Insurance Office nor CISA had assessed whether a federal insurance response was warranted. It recommended that they do so jointly and report to Congress. Both agreed.8
- 2022 to 2023. Treasury issued a request for information in 2022 (55 responses) and, at a November 2023 conference, an assistant secretary said Treasury had reached a "tentative conclusion" that any federal response should focus on catastrophic cyber risk, warning that "a poorly designed program could shift too much risk to the government."837
- March 2024. Treasury told GAO it had completed an initial assessment and decided to explore further the appropriate form of a federal cyber insurance response. As of April 2026, GAO listed both 2022 recommendations as still open, with no date for Treasury to finish or report to Congress.836
- March 2026. Treasury, in coordination with CISA, published a Federal Register request for comment on how cyber incidents are treated under the terrorism insurance program and whether changes are warranted; GAO's recommendation status page records it as a March 2026 request for comments and data.8 Analysts quoted at the time described it as exploratory, not a proposal.38
As of Nov. 15, 2026 there is no federal cyber insurance backstop and no bill to create one that this lesson found moving in Congress. The terrorism program itself is due for reauthorization by Congress before it expires on Dec. 31, 2027, which is the most likely vehicle for any change.36
Buying cyber insurance at scale
For a small business, buying cyber insurance is mostly choosing among quotes. For a large organization it is a financing decision that belongs in the enterprise risk process, with the CFO, the risk manager, the CISO and the general counsel's office at the table and a board committee overseeing it.
How much limit to buy
Brokers will show benchmarks of what similar companies buy. Benchmarks tell you what others did, not what you need. A better method starts from the organization's own loss scenarios, the quantitative risk analysis described in 108-02:
- Pick three to five plausible severe scenarios: enterprise ransomware with two weeks of downtime, theft of the largest sensitive data set, a two-day outage at the most critical cloud or software vendor, a large wire fraud.
- Estimate each scenario's loss range by component (response, notification, lost income, extra expense, liability, regulatory), using the organization's own revenue per hour of downtime, record counts and past costs.
- Map each component against the policy: covered, sublimited, excluded, below the retention.
- Decide how much of the uncovered remainder the balance sheet can absorb, which is a board risk appetite decision.
- Price additional limit and lower retentions against the expected reduction in uncovered loss. Additional limit high in the tower is usually cheaper per dollar because it is less likely to be reached.
Repeat the exercise each year, and after major changes such as an acquisition, a new critical vendor or a new line of business. Budgets, Staffing and the Business Case shows how to fold premiums and retentions into the security budget discussion; Metrics and Reporting to Leadership shows how to put the result in front of the board.
Retentions, captives and alternatives
- Higher retentions lower premiums and are rational for an organization that can absorb frequent smaller losses. They also mean the organization manages smaller incidents itself, so it needs its own retained response firms.
- Captive insurers. A captive is an insurance company owned by the organization (or a group of organizations) to insure its own risks. Large companies use captives to fund retentions, to write coverage the commercial market will not, or to access reinsurance directly. Marsh's captive practice has called cyber one of the top five growth areas for captives worldwide and cited about $166 million of cyber premium across captives it tracks, a small but growing share of the market.39 A captive needs capital, regulatory approval in its domicile, and actuarial and tax work, so it suits organizations already running one for other lines.
- Parametric cover pays a fixed amount when a defined trigger occurs, such as a cloud provider outage longer than a set number of hours, measured by an independent monitor, without proving the actual loss. It pays fast and avoids loss adjustment disputes, at the cost of basis risk: the payment may not match the real loss.
- Self-insurance, with or without a funded reserve, for organizations that judge the commercial terms poor value. Many public entities choose a pool instead (below).
The other policies in the room
A cyber event can touch several policies: directors and officers (D&O) liability for securities suits and derivative claims after a breach, property and business interruption, crime (social engineering and funds transfer fraud), kidnap and ransom, errors and omissions for technology and professional firms, and general liability. Shareholder suits over cyber disclosures usually fall under D&O, not cyber, policies, which matters now that public companies must describe cyber risk management and governance in 10-K Item 1C and disclose material incidents under Item 1.05.2840 Review the cyber exclusions and cyber grants in all of them together, with one broker or one internal owner responsible for the gaps between them.
Board oversight of the insurance program
The board, or the committee that oversees cyber risk, should see at least once a year: the limits, retentions and sublimits bought and how they compare with the loss scenarios; the main exclusions (war, infrastructure, system failure, bodily injury); any control conditions and how management verified the application; the ransom payment authority; the approved vendor list; and claims history. Directors should be able to answer the question every reporter asks after an incident: "Was this covered?" 108-04 explains the board's legal duty of oversight; insurance is one of the facts that duty touches. Public companies should also make sure any statement in SEC filings about insurance (MGM's 2023 statement that it expected its policy to cover the financial impact is an example) is accurate and kept current.17
Public entities and risk pools
Cities, counties, school districts, hospitals owned by public districts, and state agencies face the same attacks as companies, with fewer budget options and public records duties. Most do not buy cyber insurance alone. They join a risk pool: a group of public entities that share losses and buy reinsurance together, governed by elected officials of the member entities. Pools give small governments access to cover, response services and training they could not buy alone; the trade-off is that pool coverage terms are set for the group, and pools face the same pricing pressures as the commercial market when they buy reinsurance.
Idaho state agencies
Idaho's executive branch agencies are covered through the Department of Administration's Risk Management Program. The program states that all state agencies participate in core coverages including liability and "cyber liability," which it describes as coverage "from the risk posed by conducting business over the internet," funded through each agency's insurance appropriation and renewed each year.41 The program's public page does not state limits. Legislators and reporters can ask the Department of Administration for the current limits, retention, sublimits, war exclusion wording and claims history, subject to the security exemptions in the Public Records Act discussed below.
Idaho counties and local governments
The Idaho Counties Risk Management Program (ICRMP), created in 1985, describes itself as a "Public Entity and Regulated Insurer" owned by its members, governed by a board of nine local elected officials, with 1,006 members. Its multiline coverage for members includes cyber liability among the lines listed on its website, which does not publish cyber limits or terms.42 Because ICRMP files annual financial statements with the Idaho Department of Insurance, its financial condition is public.43 A county commissioner or city council member should know, for their own entity, what the pool's cyber coverage pays, what controls the pool requires or offers, and whether the entity has bought additional cover above it.
Records and public entities
Insurance applications contain a detailed map of an organization's security weaknesses. Idaho's Public Records Act exempts certain records about cybersecurity devices, programs and systems and about critical infrastructure vulnerabilities from disclosure; the critical infrastructure exemption applies only when disclosure is reasonably likely to jeopardize safety, and both exemptions leave out public expenditure records.44 The policy itself, premiums paid, the limits bought, claims paid and settlement amounts are ordinarily financial records of the public body. Agencies should mark security-sensitive parts of applications accordingly; requesters should ask for the financial terms and claims history, which rarely threaten security.45

Does insurance improve security?
Insurers say they make insureds safer by asking questions, setting conditions, offering scans and services, and paying for expert response. Skeptics argue that insurance mostly moves money and may even encourage ransom payments. The research is thinner than either side claims, but it points in a consistent direction.
- RUSI, 2023. The Royal United Services Institute, with the Universities of Kent, Oxford Brookes and De Montfort and funding from the U.K. National Cyber Security Centre, studied ransomware and cyber insurance for a year. Its occasional paper Cyber Insurance and the Ransomware Challenge (August 2023) found no strong evidence that insured victims pay ransoms more often than uninsured ones, so in the authors' view insurance is not, as many believe, directly fuelling the ransomware epidemic. It called insurance "one of the few market-based levers" for improving organizations' security, but found that insurers' influence was limited by low market take-up, weak underwriting evidence and a lack of shared learning from incidents, and said insurance should not replace minimum security standards set by law.9
- Woods and Moore. Daniel Woods and Tyler Moore, reviewing how insurers assess and price risk, concluded that cyber insurance was then a weak form of security governance: applications collected little technical detail, many pricing models gave security little or no weight, and premium discounts gave only a "marginal incentive to invest in security." They found post-incident services were the strongest part of the product.29
- Since 2021. The hard market changed underwriting. Insurers now refuse or price up organizations without MFA, EDR and offline backups, scan applicants' internet-facing systems, and send alerts to insureds about exposed vulnerabilities. Marsh McLennan's 2023 analysis tied specific controls to fewer events.20 Insurer-reported data suggest falling severity among their policyholders, but these are self-reported by firms selling the product.14
Put together: insurance has pushed a minimum set of controls across many organizations and funds expert response, but it does not set a high bar, its price signals are weak in soft markets, and it cannot be a substitute for a security program. For a large organization, the useful question is not whether insurance makes you secure, but whether the underwriting questions reveal gaps you would want to fix anyway. They usually do.
Sample: Cyber Insurance Renewal Readiness Checklist
Sample only. Adapt it to your organization's risks, systems and laws, have qualified people review it, and do not treat it as a substitute for a risk assessment.
Follows: NIST IR 8286 Rev. 1 (December 2025) on risk response and transfer; NIST Cybersecurity Framework 2.0 (Feb. 26, 2024), Govern function; the FTC's business guide to cyber insurance; OFAC's ransomware advisory of Sept. 21, 2021; New York DFS Insurance Circular Letter No. 2 (2021); SEC Release 33-11216 (2023); and the case law described in this lesson.
1. Ownership and calendar
- [ ] Renewal owner named: [Risk Manager]; security owner: [CISO]; wording reviewer: [general counsel's office or compliance officer]; approver: [CFO]; board committee: [Risk / Audit Committee].
- [ ] Renewal kickoff at least [120] days before expiration on [date]; application to the CISO's office by [date]; board committee briefing by [date].
- [ ] Broker's scope of work and compensation disclosed in writing.
2. Loss scenarios and limits
- [ ] [3 to 5] severe scenarios updated with current revenue per hour of downtime, record counts and vendor dependencies.
- [ ] Each scenario mapped against current limits, sublimits, retentions, waiting periods and exclusions; uncovered remainder stated in dollars.
- [ ] Board-approved risk appetite for uncovered cyber loss: [$ amount].
- [ ] Options priced: higher limit, lower retention, removal of ransomware sublimit or coinsurance, dependent business interruption for named vendors, system failure cover.
3. Application and evidence
- [ ] Every control question assigned to a named owner with evidence attached: MFA coverage report (remote access, email, privileged accounts, vendors, backups); EDR coverage percentage and monitoring arrangement; most recent backup restore test; privileged account inventory; patch aging and KEV status; exposed services list; end-of-life systems and isolation; incident response plan and last exercise date.
- [ ] Partial answers stated as partial, with exceptions and remediation dates.
- [ ] Second-line or internal audit sample of evidence completed by [date].
- [ ] Signing officer briefed with a written verification memo.
- [ ] Final application, supplementals, scans and evidence archived with the policy.
4. Wording review
- [ ] War and state-backed attack exclusion: clause name and version; attribution method; collateral damage; essential services definition; same clause across all layers.
- [ ] Infrastructure, system failure and bodily injury exclusions compared with property, general liability and crime policies; no gaps between policies.
- [ ] Knowledge group, severability and non-rescission language.
- [ ] Retroactive date unchanged or matched if changing insurers.
- [ ] Notice requirements for each layer; definition of "claim" and "incident."
- [ ] Regulatory fines and PCI assessments: sublimits and insurability language.
- [ ] Defense costs inside or outside limits; consent to settle.
- [ ] Dispute resolution, governing law and suit limitation period.
5. Response readiness
- [ ] Organization's own response firm and law firm pre-approved by the insurer in writing, with rates: [names].
- [ ] Insurer hotline and policy numbers for every layer in the incident response plan and on the crisis team's call cards.
- [ ] Ransom payment authority documented: [CEO with Board Chair / Committee]; OFAC screening and law enforcement reporting steps included.
- [ ] Cost tracking codes and downtime logging procedure ready for business interruption claims.
- [ ] Tabletop exercise in the last 12 months included the insurer notice and consent steps.
6. Reporting
- [ ] Board committee briefed on limits, key exclusions, conditions, verification, payment authority and claims history.
- [ ] Statements about insurance in SEC filings, bond disclosures or public reports reviewed for accuracy.
- [ ] Open remediation items from the application tracked in the risk register with owners and dates.
Some of the items above, such as closing exposed remote access or extending MFA to vendors before renewal, are changes to production systems; plan each change, test it and keep a way back.
Pictures, Numbers and Infographics
| Number | What it is | Source |
|---|---|---|
| About $9.46 billion | U.S. cyber direct written premium in 2025, including alien surplus lines, up about 3.5 percent | NAIC 2026 report2 |
| About $7.46 billion | Premium written by U.S.-domiciled insurers in 2025 (about $2.75 billion in 2020) | NAIC 2026 and 2025 reports210 |
| About 7 percent | Drop in U.S. cyber premium from 2023 to 2024, the market's first decline | NAIC 2025 report10 |
| 53 | U.S. cyber loss ratio (losses plus defense and cost containment) in 2025, up from 48.7 in 2024 | NAIC 2026 report2 |
| 56,323 | Claims reported to U.S. domestic insurers in 2025, up 20 percent | NAIC 2026 report2 |
| About 5.84 million | Domestic cyber policies in force in 2025, up 33 percent; about 4 percent of premium is endorsements | NAIC 2026 report2 |
| +130 percent | U.S. cyber price change in the fourth quarter of 2021 (U.K. +92 percent) | Marsh Global Insurance Market Index, Q4 202113 |
| Flat; global down 4 percent | U.S. and global cyber rate change in the third quarter of 2026; 13th straight quarterly global decline | Marsh, Oct. 21, 20263 |
| 58 percent; $269,000 | Share of Coalition's 2025 claims from email compromise and funds transfer fraud; average ransomware loss | Coalition 2026 Cyber Claims Report14 |
| 86 percent | Share of Coalition policyholders receiving ransom demands in 2025 that refused to pay | Coalition 2026 Cyber Claims Report14 |
| Aug. 26, 2022 | Date Travelers and International Control Services stipulated that a cyber policy running April 4, 2022 to April 4, 2023 was void from inception over MFA answers | Insurance Journal, Aug. 30, 20224 |
| $1.75 billion over $150 million | Merck's property program limits and deductible in the NotPetya dispute; about $700 million contested at the end | Insurance Journal, Jan. 5, 202422 |
| About $100 million | Mondelez's NotPetya property claim, settled October 2022 | The Record, Oct. 31, 202221 |
| March 31, 2023 | Date from which Lloyd's syndicates must include state-backed cyber attack exclusions in standalone cyber policies | Lloyd's Market Bulletin Y53816 |
| 8.5 million | Windows devices Microsoft estimated were affected by the July 19, 2024 CrowdStrike update | Microsoft, July 20, 202430 |
| $400 million to $1.5 billion | CyberCube's estimate of insured losses from the CrowdStrike outage ($300 million to $1 billion per Guy Carpenter) | Insurance Journal, July 25, 2024; Guy Carpenter, Aug. 1, 2024731 |
| About £1.9 billion | Cyber Monitoring Centre estimate of U.K. economic damage from the 2025 Jaguar Land Rover attack | Reinsurance News, October 202534 |
| 2 | Open GAO recommendations (to Treasury and CISA) to assess a federal insurance response to catastrophic cyber incidents, made June 21, 2022 | GAO-22-1042568 |
| Sept. 21, 2021 | OFAC's updated ransomware payment advisory: strict liability for sanctions violations | OFAC advisory26 |
Common mistakes
Signing an application nobody checked
Travelers alleged that International Control Services' application said multifactor sign-in was in place when it was not, and the company agreed to have the policy declared void from inception, leaving it with no coverage for the ransomware event.4 Fix: verify every control answer with evidence, state exceptions, and brief the signer in writing.
Relying on a policy that never mentions cyber
Mondelez had no standalone cyber policy and sought about $100 million under its property policy after NotPetya; it spent four years in litigation over a war exclusion before settling on undisclosed terms.21 Merck's property claim took more than six years from the attack to a settlement.22 Both eventually did better than their insurers' first position, but years of uncertainty is a cost. Fix: buy affirmative cyber cover and make every other policy say clearly whether it covers or excludes cyber losses, as New York DFS has told its insurers to do.25
Leaving cover unfinished
According to a trade press report that the company has not confirmed, Jaguar Land Rover was attacked in 2025 while its cyber placement was still being negotiated, so no policy responded to a shutdown the Cyber Monitoring Centre valued at about £1.9 billion across the U.K. economy.3534 Fix: do not let a lapse or a long negotiation leave the organization bare; extend expiring cover while terms are settled.
Assuming a vendor outage is covered
Estimates after the July 2024 CrowdStrike outage put insured losses at a fraction of the economic loss, in part because many policies covered dependent business interruption only for security failures, not vendor errors, or only after long waiting periods.732 Fix: list the critical vendors, model a multi-day outage at each, and buy system failure and dependent business interruption cover sized to it.
Leaving the control gap the application asks about
The Change Healthcare attack began, according to UnitedHealth's chief executive's Senate testimony, with stolen credentials used on a remote access portal without multifactor authentication.19 That is the first question on most cyber applications. Fix: treat the application's control list as a minimum, test coverage across subsidiaries and acquired companies, and fix gaps before renewal.
Deciding about ransom on the night
OFAC holds payers strictly liable for paying sanctioned parties, and names insurers and response firms among facilitators.26 Organizations that have not decided in advance who may approve a payment, what facts are required and how law enforcement and sanctions screening fit in make the decision under the worst possible conditions. Fix: write the authority and the steps into the incident response plan and exercise them.
Reading only the primary policy
Excess layers may add their own exclusions or notice rules, and some large claims have turned on which layer had which wording. Fix: one person reviews the whole tower and all related policies each year, with the differences listed for the board committee.
Questions people ask
Is cyber insurance required by law?
Generally no. No federal law requires private organizations to buy it. Contracts often do: customers, lenders, landlords and government agencies may require vendors to carry cyber or technology errors and omissions coverage at stated limits. Public entities may be required by their own rules or pool agreements. Check your contracts, and see 108-08 on requiring insurance from your own vendors.
Will the policy pay a ransom?
Many policies will reimburse a lawful ransom payment made with the insurer's consent, within the extortion sublimit and any coinsurance. Neither the policy nor the insurer can make a payment to a sanctioned party lawful, and OFAC, New York DFS, CISA and the FBI all discourage paying.2625 The decision stays with the organization.
Are regulatory fines covered?
Sometimes. Policies typically cover fines and penalties "where insurable by law," and the FTC's guide lists fees, fines and penalties among possible first-party costs.1 Whether a given fine is insurable depends on the law that applies and on the type of penalty; some, such as many criminal penalties, are not insurable anywhere. Defense costs for investigations are more reliably covered than the penalties themselves.
Does a SOC 2 report or ISO certificate lower the premium?
It can help the underwriting conversation, but insurers focus on specific controls, such as MFA breadth, EDR coverage, backup testing and patch speed, more than on certificates. Bring the evidence for those controls. SOC Reports and Attestations and Certifications and Government Authorizations explain what those documents do and do not show.
What happens to insurance when we buy a company?
The target's policy usually covers its pre-closing incidents only if notice is given under that policy, and many policies end or change on a change of control. The buyer's policy may not cover the target's prior acts. Diligence should collect the target's policies, applications and claims history, and the deal should address run-off cover. Security in Mergers and Acquisitions covers this in detail.
Who should own cyber insurance inside the organization?
Usually the risk manager or CFO's office owns the purchase, the CISO owns the control answers and evidence, the general counsel's office or compliance officer reviews the wording, and a board committee oversees the result. When one person tries to do all four, the application and the wording are where it breaks.
Why did our premium fall when attacks are rising?
Because price follows the supply of insurance capital and insurers' recent profits as much as current risk. After the 2021 to 2022 increases, cyber was profitable, new capacity arrived, and rates fell for many quarters even as claims rose.23 Use the soft market to improve terms, not only to cut cost; the cycle will turn.
Hard questions
Should insurance be allowed to pay ransoms?
For allowing it: Some victims, such as hospitals with patients at risk or companies whose backups failed, have no realistic alternative, and insurance lets them recover without bankruptcy. Banning reimbursement would push payments underground, punish victims rather than criminals and reduce reporting. The 2023 RUSI study found no strong evidence that insured victims pay more often than uninsured ones and did not recommend a ban.9
For restricting it: Every payment funds the next attack. Regulators have said so plainly: New York DFS "recommends against making ransom payments," and OFAC "strongly discourages" them.2526 If insurance reliably pays, criminals can price demands to the policy.
What the evidence shows: The research available does not show insurance causing payments, and payment rates have fallen as backups and response improved; Coalition reported 86 percent of its policyholders refusing to pay in 2025.14 The policy tools that most regulators have chosen are sanctions screening, law enforcement reporting and insurer consent, not a ban.
Should the federal government backstop catastrophic cyber losses?
For: A truly systemic event, such as a destructive attack on a dominant cloud provider or the power grid, could exceed what private insurers can pay, and they already exclude much of it. A backstop with clear rules could keep cover available and, as GAO noted, could require security standards and contributions from market participants.8
Against: A backstop could shift too much risk to taxpayers and weaken incentives to secure systems, as Treasury's own officials warned in 2023.37 Defining and attributing a catastrophic cyber event is much harder than defining a hurricane, and a program could become a subsidy for firms that underinvest.
What the evidence shows: After four years of study, Treasury and CISA have not finished their joint assessment, GAO's recommendations remain open, and the 2026 request for comment was exploratory.838 The CrowdStrike and Jaguar Land Rover events showed large uninsured losses but did not break the insurance market. The decision is a policy choice that Congress has not yet taken up.
Is it fair for insurers to exclude state-backed attacks?
For the insurers: State-backed operations can hit thousands of organizations at once, as NotPetya did. Lloyd's has said such losses could "greatly exceed what the insurance market is able to absorb," and clear exclusions let insurers keep offering cover for everything else.23
For policyholders: Victims cannot control who attacks them, attribution is slow and political, and an exclusion turning on a government's statement can leave a hospital or manufacturer bare for an event it could not prevent. The Merck court refused to stretch an old war exclusion to a cyberattack with no link to military action.5
What the evidence shows: The market has moved to explicit exclusions with stated thresholds and attribution methods, which is clearer than the old language. Whether they are fair depends on the thresholds; buyers with negotiating power have won narrower versions. The unresolved risk is the event between ordinary crime and war, which neither private insurance nor any federal program clearly covers.
Should public bodies buy commercial cyber insurance, join pools or self-insure?
For commercial insurance or pools: Small governments cannot absorb a multi-million-dollar incident or organize expert response on their own. Idaho's state agencies and ICRMP's members already get cyber cover through pooled programs, which also bring training and response services.4142
For self-insurance: Large states can spread risk across many agencies and years, and premiums include insurer expenses and profit. Money spent on premiums is money not spent on staff and controls.
What the evidence shows: There is little public data comparing outcomes. What is clear is that whatever the structure, taxpayers pay either way, so legislators should see the limits, retentions, exclusions and claims history every year.
For reporters and advocates
What public records exist.
- SEC filings on EDGAR. Form 8-K Item 1.05 incident filings and later 10-Q and 10-K filings often say whether a company expects insurance to cover costs and later report insurance recoveries; 10-K Item 1C describes cyber risk management. EDGAR full-text search finds phrases such as "cyber insurance" or "insurance recoveries" in a date range.4640 Merck's 2017 10-K, which reported insurance recoveries of about $45 million against NotPetya costs, is an example.16
- NAIC market data. The NAIC's annual cybersecurity insurance market report gives premiums, policies, loss ratios and the largest insurers by name.2 State insurance departments publish insurers' annual statements, including those of public entity pools such as ICRMP in Idaho.43
- Court records. Coverage disputes such as Travelers v. International Control Services (C.D. Ill. No. 22-cv-2145) and the Merck and Mondelez cases produce pleadings, policies and exhibits; federal dockets are on PACER, with many mirrored free on CourtListener, and state appellate opinions are on court websites.45
- Federal reports. GAO's reports on cyber insurance and terrorism insurance, Treasury's Federal Insurance Office annual reports and Federal Register notices, OFAC advisories and enforcement actions.83626
- Public entity records. In Idaho, requests under the Public Records Act (Idaho Code Title 74, Chapter 1) can seek a state agency's or local government's cyber policy declarations, premiums, pool contribution, limits and claims paid; security-sensitive parts of applications may be withheld under the cybersecurity and critical infrastructure exemptions.4544
What to ask an organization after an incident. Did you have cyber insurance, and what were the limit and the retention? Was there a ransomware sublimit or coinsurance? Has the insurer accepted coverage, reserved its rights or denied? Did the insurer require particular response firms? Who approved any ransom payment, was it screened for sanctions, and was law enforcement told? Who verified the security answers on the last application, and were they accurate? How much of the total cost do you expect to be uninsured?
What to ask a public body before an incident. What cyber coverage do you have, through which program or pool, with what limits, retention and exclusions? What did it cost this year and five years ago? What controls did the insurer or pool require, and do you meet them? Has the board or commission been briefed on the coverage? Who can approve a ransom payment?
Red flags. An organization that cannot say whether it was insured; an application that claimed controls the incident showed were missing; statements that "insurance will cover it" before the insurer has accepted the claim; a public body with no cyber cover and no reserve; and a company whose SEC filings promise strong security while its insurance terms show sublimits and conditions it has not met.
The Idaho angle. Idaho's state agencies carry cyber liability through the Department of Administration's Risk Management Program, and Idaho counties and many local entities through ICRMP. Neither publishes cyber limits or claims on its website. Legislators reviewing agency budgets, county commissioners approving pool contributions, and reporters covering incidents at Idaho hospitals, schools and local governments can ask for those numbers and for how each program verifies its members' security.4142
Where to go next
- NAIC cybersecurity insurance market reports: Annual U.S. market data from state insurance regulators: premiums, policies, loss ratios, claims and the largest insurers. Free. Use it to check broker claims about the market and to see who writes cyber cover.
- CISA #StopRansomware Guide: Free federal guide with prevention practices and a response checklist. Use it to test the controls underwriters ask about and to build the response steps the policy plugs into.
- OFAC ransomware advisory: Treasury's 2021 advisory on sanctions risk in ransom payments. Free. Read it before writing ransom payment authority into the incident response plan.
- FTC: Cyber Insurance: Free plain-language list of first-party and third-party coverages and questions to ask an insurer; written for small businesses but a useful checklist for any buyer.
- GAO-22-104256, Cyber Insurance: Free GAO report on catastrophic cyber risk, exclusions and the federal backstop question, with recommendation status updates. For policymakers, legislative staff and reporters.
- New York DFS Cyber Insurance Risk Framework: Free regulator guidance to insurers on silent cyber, systemic risk, underwriting and ransom payments. Useful for buyers to understand what insurers are expected to ask.
- Idaho Risk Management Program: The Idaho Department of Administration's page on state agency insurance, including cyber liability. Free. A starting point for Idaho legislators and reporters asking about state coverage.
Still being checked (10 items)
These points could not be fully confirmed against a primary source when this lesson was written and fact checked in November 2026. The text above states each one as uncertain.
- The text of the Merck appellate opinion (May 1, 2023) was not retrieved; quotes come from a law firm summary and news reports.
- The Travelers v. ICS stipulation was not read directly; facts come from Insurance Journal's Aug. 30, 2022 report. The policy limit was not confirmed and is not stated.
- The Lloyd's bulletin Y5381 date is reported as Aug. 16, 2022 by Insurance Journal; the bulletin PDF read may show Aug. 17, 2022, so the lesson says August 2022.
- The original publication date of LMA5564 to LMA5567 (reported elsewhere as November 2021) was not confirmed and is not stated.
- Press reports that Jaguar Land Rover had no cyber policy in place were not confirmed by the company.
- CrowdStrike insured loss figures are early model estimates; no final industry count of paid claims was found.
- The March 2026 Treasury request for comment is confirmed by GAO's GAO-22-104256 recommendation status page (April 2026 update); the Federal Register notice itself was not retrieved, and its comment period and any outcome are not stated.
- Whether any Idaho statute bars public agencies from paying ransoms: none was found, but a full search of the Idaho Code was not done.
- ICRMP's cyber limits and terms and the State of Idaho's cyber limits are not published on the pages read.
- The captive cyber premium figure of about $166 million is from a trade article quoting Marsh; its year is not stated.
Last verified November 15, 2026. General education, not legal, compliance, audit or other professional advice; rule versions and dates are stated in the text, and the sources used are listed in the Notes and Sources below. Check the regulator's current text before relying on any point.
Notes
- Federal Trade Commission, "Cyber Insurance," Cybersecurity for Small Business guidance (lists of first-party coverage, including "Fees, fines, and penalties related to the cyber incident," and third-party coverage, and questions to ask an insurer) (read Nov. 15, 2026). Read it online. ↩
- National Association of Insurance Commissioners, 2026 Report on the Cybersecurity Insurance Market (2025 data: about $9.46 billion U.S. direct written premium including alien surplus lines, up about 3.5 percent; domestic $7.46 billion; 5.84 million domestic policies in force, up 33 percent; primary about 64 percent, excess about 32 percent and endorsements about 4 percent of premium; loss ratio with defense and cost containment 53, up from 48.7; 56,323 claims, up 20 percent; large-business average severity about $4.4 million, up from about $2.2 million; top groups Chubb, Beazley and Travelers; top 20 about 76.8 percent; first quarter of 2026 the "eighth consecutive quarter of U.S. cyber rate cuts," citing AM Best) (read Nov. 15, 2026). Read it online. ↩
- Marsh, "Global commercial insurance rates fall 5% in Q3 2026," Global Insurance Market Index press release, Oct. 21, 2026 (global cyber rates down 4 percent, "the thirteenth consecutive quarter of declines"; U.S. cyber rates flat "after an extended period of declines") (read Nov. 15, 2026). Read it online. ↩
- Insurance Journal, "Travelers, Insured Agree to Rescind Cyber Policy Over MFA Misrepresentation" (news report), Aug. 30, 2022, describing Travelers Property Casualty Co. of America v. International Control Services, Inc., No. 22-cv-2145 (C.D. Ill.): policy term April 4, 2022 to April 4, 2023; application and MFA attestation signed in March 2022; ransomware event in May 2022; joint stipulation filed Aug. 26, 2022 declaring the policy "null and void, from its inception" and dismissing with prejudice. Secondary source; the stipulation was not read directly for this lesson (read Nov. 15, 2026). Read it online. ↩
- DAC Beachcroft, "Merck prevails on war clause appeal," May 15, 2023, summarizing Merck & Co. v. ACE American Insurance Co., N.J. Super. Ct. App. Div., Nos. A-1879-21 and A-1882-21, decided May 1, 2023 (opinion by Judge Currier: the exclusion "requires the involvement of military action" and covers "actions clearly connected to war or, at least, to a military action or objective"). Secondary source; the opinion itself was not retrieved for this lesson (read Nov. 15, 2026). Read it online. ↩
- Lloyd's, Market Bulletin Y5381, "State backed cyber-attack exclusions," August 2022, Tony Chaudhry, Underwriting Director (requirements apply at inception or renewal of each standalone cyber-attack policy from March 31, 2023; minimum requirements quoted in the text) (read Nov. 15, 2026). Read it online. ↩
- Insurance Journal, "Insured Losses From CrowdStrike Outage Could Reach $1.5B, CyberCube Says" (Reuters report), July 25, 2024 (CyberCube estimate $400 million to $1.5 billion; Parametrix estimate of insured losses of $540 million to $1.08 billion for Fortune 500 companies excluding Microsoft) (read Nov. 15, 2026). Read it online. ↩
- U.S. Government Accountability Office, Cyber Insurance: Action Needed to Assess Potential Federal Response to Catastrophic Attacks, GAO-22-104256, June 21, 2022, with recommendation status as updated through April 2026 (Treasury's 2022 request for information drew 55 responses; Treasury completed an initial assessment in March 2024; both recommendations open) (read Nov. 15, 2026). Read it online. ↩
- University of Kent, "Cyber insurance is not fuelling the ransomware epidemic," Aug. 1, 2023, announcing the RUSI occasional paper Cyber Insurance and the Ransomware Challenge by RUSI with the Universities of Kent, Oxford Brookes and De Montfort, funded by the U.K. National Cyber Security Centre (quotes from Dr. Jason Nurse and the paper's findings as summarized by the university) (read Nov. 15, 2026). Read it online. ↩
- National Association of Insurance Commissioners, 2025 Cybersecurity Insurance Report (2024 data: total direct written premium about $9.14 billion, the market's "first ever reduction," about 7 percent below about $9.84 billion in 2023; domestic premium by year 2020 to 2024; policies in force; nearly 50,000 claims, up almost 40 percent) (read Nov. 15, 2026). Read it online. ↩
- NIST, Integrating Cybersecurity and Enterprise Risk Management (ERM), NIST IR 8286 Revision 1, December 2025, Sec. 3.5 and Table 2 (risk response types, including transfer) (read Nov. 15, 2026). Read it online. ↩
- Insurance Journal, "Cyber Direct Premiums Are Up but Majority of Policies Are Endorsements: NAIC," Nov. 9, 2026 (domestic premium nearly tripled from $2.75 billion in 2020; surplus lines insurers write nearly two-thirds of U.S. cyber premium). Secondary source (read Nov. 15, 2026). Read it online. ↩
- Insurance Journal, "Global Commercial Insurance Rates Keep Moderating, With Exception of Cyber: Marsh," Feb. 2, 2022, reporting Marsh's Global Insurance Market Index for Q4 2021 (U.S. cyber prices up 130 percent, up from 96 percent in Q3; U.K. up 92 percent; tied to ransomware claims) (read Nov. 15, 2026). Read it online. ↩
- Insurance-Canada.ca, "Coalition's 2026 Cyber Claims Report Finds Initial Ransom Demands Surged 47%, But Most Businesses Refuse to Pay," March 6, 2026, reporting Coalition's release of March 5, 2026 (full-year 2025 data from more than 100,000 policyholders in five countries: claims frequency up 3 percent, severity down 19 percent to an average $116,000; BEC and funds transfer fraud 58 percent of incidents; ransomware average loss $269,000; 86 percent refused to pay). Insurer's own data (read Nov. 15, 2026). Read it online. ↩
- FBI, Internet Crime Complaint Center, 2024 IC3 Annual Report (859,532 complaints; $16.6 billion in reported losses; business email compromise losses $2,770,151,146) (read Nov. 15, 2026). Read it online. ↩
- Merck & Co., Inc., Annual Report on Form 10-K for the fiscal year ended Dec. 31, 2017 (SEC EDGAR), discussion of the June 27, 2017 cyber-attack: sales effect of about $260 million in 2017 and expenses of $285 million, "net of insurance recoveries of approximately $45 million" (read Nov. 15, 2026). Read it online. ↩
- MGM Resorts International, Current Report on Form 8-K, filed Oct. 5, 2023 (negative impact of about $100 million on Adjusted Property EBITDAR; belief that cyber insurance would be sufficient to cover the financial impact) (read Nov. 15, 2026). Read it online. ↩
- Cybersecurity and Infrastructure Security Agency, Known Exploited Vulnerabilities Catalog, continuously updated (read Nov. 15, 2026). Read it online. ↩
- U.S. Senate Committee on Finance, hearing, Hacking America's Health Care: Assessing the Change Healthcare Cyber Attack and What's Next, May 1, 2024, S. Hrg. 118-785 (written testimony of UnitedHealth Group chief executive Andrew Witty: compromised credentials used on a Citrix remote access portal that did not have multifactor authentication) (read Nov. 15, 2026). Read it online. ↩
- Marsh McLennan, "Groundbreaking Research from Marsh McLennan Reveals Direct Link between Key Cybersecurity Controls and Reduced Cyber Risk," press release, April 6, 2023 (claims data matched to Cybersecurity Self-Assessment answers; automated hardening strongest "by a wide margin"; patching high-severity vulnerabilities within seven days, implemented by 24 percent of organizations studied; MFA effective when broadly deployed). Broker's own study (read Nov. 15, 2026). Read it online. ↩
- The Record, "Mondelez and Zurich reach settlement in NotPetya cyberattack insurance suit," Oct. 31, 2022 (claim of about $100 million under a property policy; "hostile or warlike action" exclusion; terms not disclosed). News report (read Nov. 15, 2026). Read it online. ↩
- Insurance Journal, "Merck Settles Coverage Dispute With Insurers Over War Exclusion in NotPetya Attack" (Bloomberg report), Jan. 5, 2024 (trial ruling January 2022; appellate affirmance 2023; New Jersey Supreme Court review granted July 2023; $1.75 billion in property limits above a $150 million deductible; about $700 million disputed by the eight remaining insurers; confidential settlement) (read Nov. 15, 2026). Read it online. ↩
- Insurance Journal, "Lloyd's Insurers Must Exclude Catastrophic State-Backed Cyber-Attacks," Aug. 19, 2022 (bulletin published Aug. 16, 2022; losses could "greatly exceed what the insurance market is able to absorb"; effective March 31, 2023) (read Nov. 15, 2026). Read it online. ↩
- DAC Beachcroft, "War exclusions in cyber policies: an overview" (LMA5564 to LMA5567 and their A and B versions; thresholds such as "major detrimental impact" and widespread events in market clauses). Secondary source (read Nov. 15, 2026). Read it online. ↩
- New York State Department of Financial Services, Insurance Circular Letter No. 2 (2021), "Cyber Insurance Risk Framework," Feb. 4, 2021 (read Nov. 15, 2026). Read it online. ↩
- U.S. Department of the Treasury, Office of Foreign Assets Control, Updated Advisory on Potential Sanctions Risks for Facilitating Ransomware Payments, Sept. 21, 2021 (superseding the Oct. 1, 2020 advisory) (read Nov. 15, 2026). Read it online. ↩
- CISA, MS-ISAC, NSA and FBI, #StopRansomware Guide (prevention practices and response checklist; free) (read Nov. 15, 2026). Read it online. ↩
- U.S. Securities and Exchange Commission, "Cybersecurity Risk Management, Strategy, Governance, and Incident Disclosure," Release Nos. 33-11216 and 34-97989, July 26, 2023, 88 Fed. Reg. 51896 (Aug. 4, 2023) (Form 8-K Item 1.05; Regulation S-K Item 106; Form 10-K Item 1C) (read Nov. 15, 2026). Read it online. ↩
- Daniel W. Woods and Tyler Moore, "Does Insurance Have a Future in Governing Cybersecurity?," IEEE Security & Privacy (2020), author's accepted manuscript (pre-underwriting assessment, pricing and post-incident services; discounts provide a "marginal incentive to invest in security"; conflicts when insurers select response providers) (read Nov. 15, 2026). Read it online. ↩
- David Weston, Microsoft, "Helping our customers through the CrowdStrike outage," July 20, 2024 ("CrowdStrike's update affected 8.5 million Windows devices, or less than one percent of all Windows machines") (read Nov. 15, 2026). Read it online. ↩
- Guy Carpenter, analysis of the July 19, 2024 global IT outage, updated Aug. 1, 2024 (likely insured loss "between $300 million and $1 billion") (read Nov. 15, 2026). Read it online. ↩
- Claims Journal, "Billions in Damages From CrowdStrike Outage to Go Uninsured" (Bloomberg report), Aug. 5, 2024 (Parametrix: about $5.4 billion in damages to Fortune 500 companies; Guy Carpenter and CyberCube insured loss estimates) (read Nov. 15, 2026). Read it online. ↩
- Insurance Journal, "Jaguar Land Rover Bailout Shows Rising Cost of Cybercrime" (Bloomberg report), Oct. 7, 2025 (attack detected Aug. 31, 2025; phased restart Oct. 8, 2025; U.K. government guarantee of a £1.5 billion loan) (read Nov. 15, 2026). Read it online. ↩
- Reinsurance News, "Jaguar Land Rover attack most economically damaging cyber event to hit UK: CMC," October 2025, reporting the Cyber Monitoring Centre's estimate of about £1.9 billion (modeled range £1.6 billion to £2.1 billion), more than 5,000 organizations affected, category 3 systemic event (read Nov. 15, 2026). Read it online. ↩
- Insurance Business, "Did Jaguar Land Rover's cyber insurance gap help pave the way to 4,000 job losses?" (trade press report that a cyber placement through the company's broker had not been completed before the attack; not confirmed by the company in any source found for this lesson) (read Nov. 15, 2026). Read it online. ↩
- U.S. Government Accountability Office, Terrorism Risk Insurance Act: Considerations for Reauthorization, GAO-25-108748, Sept. 23, 2025 (cyber attacks and the program's certification requirements; Treasury's March 2024 initial cyber assessment; status as of April and May 2025) (read Nov. 15, 2026). Expiration date: NAIC, "Terrorism Risk Insurance Act" topic page ("The current reauthorization is slated to expire December 31, 2027"), read Nov. 15, 2026. Read it online. ↩
- Claims Journal, report on Treasury Assistant Secretary Graham Steele's remarks at the Nov. 17, 2023 conference of NYU Stern's Volatility and Risk Institute and the Federal Insurance Office, Nov. 28, 2023 ("tentative conclusion" to focus on catastrophic cyber risk; "a poorly designed program could shift too much risk to the government") (read Nov. 15, 2026). Read it online. ↩
- Chris Riotta, "US Treasury Weighs Cyber Insurance Backstop," GovInfoSecurity, March 27, 2026 (Treasury Federal Register request for comment, with CISA, on cyber incidents under the Terrorism Risk Insurance Program; analysts' views). News report; the Federal Register notice itself was not retrieved for this lesson (read Nov. 15, 2026). Read it online. ↩
- Alex Wright, "Captive Insurance Expands Role in Cyber-Risk Financing Strategies," Captive.com, May 6, 2026 (Marsh Captive Solutions: cyber among the top five growth areas for captives; about $166 million of cyber premium across captives, year not stated). Trade press (read Nov. 15, 2026). Read it online. ↩
- 17 CFR 229.106 (Regulation S-K Item 106, Cybersecurity), eCFR current text (read Nov. 15, 2026). Read it online. ↩
- Idaho Department of Administration, Risk Management Program, "Insurance Coverages" (all state agencies participate in core coverages including cyber liability; funded through agency insurance appropriations; renewed annually) (read Nov. 15, 2026). Read it online. ↩
- Idaho Association of Counties, "Liability Protection and Risk Management" page describing the Idaho Counties Risk Management Program (created 1985; "Public Entity and Regulated Insurer"; 1,006 members; board of nine local elected officials; multiline coverage listing cyber liability) (read Nov. 15, 2026). Read it online. ↩
- Idaho Department of Insurance, Idaho Counties Risk Management Program annual statement for the year ended 2024, posted with insurer financial statements (read Nov. 15, 2026). Read it online. ↩
- Idaho Code 74-105, Records exempt from disclosure, including subsection (4)(b) (critical infrastructure records, including vulnerability assessments) and subsection (21) (records on cybersecurity devices, programs or systems), as amended through the 2026 session (read Nov. 15, 2026). Read it online. ↩
- Idaho Code Title 74, Chapter 1, Public Records Act, as amended through the 2026 session (read Nov. 15, 2026). Read it online. ↩
- U.S. Securities and Exchange Commission, EDGAR full-text search (read Nov. 15, 2026). Read it online. ↩
Check yourself
1. According to the NAIC's 2026 report, about how much U.S. cyber direct written premium (including alien surplus lines) was written in 2025?
The NAIC 2026 Report on the Cybersecurity Insurance Market put 2025 premium at about $9.46 billion, up about 3.5 percent from 2024. $2.75 billion was domestic premium in 2020; 5.84 million is the number of policies; $16.6 billion is the FBI IC3 2024 loss figure.
2. What happened in Travelers v. International Control Services (C.D. Ill. 2022)?
On Aug. 26, 2022 the parties filed a joint stipulation declaring the policy null and void from its inception and dismissing the case, after Travelers alleged misrepresentation of MFA use (Insurance Journal, Aug. 30, 2022).
3. What did the New Jersey Appellate Division hold in Merck v. ACE American on May 1, 2023?
The court held the exclusion "requires the involvement of military action" and affirmed the trial court's ruling for Merck (Nos. A-1879-21 and A-1882-21).
4. From what date did Lloyd's require state-backed cyber attack exclusions in standalone cyber policies?
Lloyd's Market Bulletin Y5381 (August 2022) applies at inception or renewal of each policy from March 31, 2023.
5. Under OFAC's Sept. 21, 2021 advisory, what is the basis for civil penalties for paying a sanctioned ransomware actor?
The advisory says OFAC may impose civil penalties for sanctions violations based on strict liability, and treats prompt reporting to law enforcement as a significant mitigating factor.
6. What is a ransomware coinsurance provision?
Coinsurance keeps the insured sharing a percentage of ransomware losses, a term widely added during the 2021 to 2022 hard market, as described in the lesson.
7. What range of insured losses did CyberCube estimate for the July 19, 2024 CrowdStrike outage?
CyberCube estimated $400 million to $1.5 billion (Insurance Journal, July 25, 2024); Guy Carpenter estimated $300 million to $1 billion; $5.4 billion was Parametrix's estimate of total Fortune 500 losses.
8. What did GAO recommend in GAO-22-104256 (June 21, 2022)?
GAO recommended a joint FIO and CISA assessment reported to Congress; both agencies agreed and the recommendations remained open as of April 2026.
9. What does New York DFS Insurance Circular Letter No. 2 (2021) say about silent cyber risk?
The Cyber Insurance Risk Framework (Feb. 4, 2021) tells insurers to manage and eliminate exposure to silent cyber insurance risk and to require notice to law enforcement.
10. How are Idaho's executive branch state agencies covered for cyber liability?
The Idaho Department of Administration's Risk Management Program states that all state agencies participate in core coverages including cyber liability; ICRMP serves counties and other local entities.
Terms in this lesson
- First-party coverage
- Insurance that pays the insured organization's own costs after a cyber event, such as response experts, notification, data restoration, business interruption and extortion.
- Third-party coverage
- Insurance that pays defense costs and damages when others, such as customers, business partners or regulators, make claims against the insured.
- Sublimit
- A cap for a particular coverage inside the policy's overall limit, such as a ransomware or dependent business interruption sublimit; it is part of the limit, not added to it.
- Retention
- The amount the insured pays before the insurer pays, the cyber equivalent of a deductible.
- Waiting period
- The number of hours a system outage must last before business interruption coverage begins to count lost income.
- Tower
- A program of a primary policy and excess policies stacked above it, each paying only after the layers below are used up.
- Dependent business interruption
- Coverage for lost income when an outage at a vendor or cloud provider, rather than the insured's own systems, stops the insured's business.
- Rescission
- An insurer's treatment of a policy as void from the start, usually because of a material misrepresentation on the application.
- Silent cyber
- Cyber losses that might be covered by a policy that does not mention cyber, such as an older property policy; also called non-affirmative cyber risk.
- State-backed cyber attack exclusion
- A clause excluding losses from cyber operations carried out by or for a state, required by Lloyd's for standalone cyber policies from March 31, 2023, usually with an attribution method and impact thresholds.
- Panel vendor
- A law firm, forensic firm, notification vendor or other provider pre-approved by the insurer, usually at agreed rates.
- Loss ratio
- Losses (here including defense and cost containment expenses) divided by premium; a loss ratio of 53 means about $53 of claims per $100 of premium.
- Captive insurer
- An insurance company owned by the organization or a group of organizations to insure their own risks.
- Parametric cover
- Insurance that pays a fixed amount when a defined, independently measured trigger occurs, such as a cloud outage of set length, without proof of actual loss.
- Risk pool
- A group of public entities that share losses and buy reinsurance together, governed by officials of the members.
Sources
- NAIC, 2026 Report on the Cybersecurity Insurance Market (2025 data) (2026) checked 2026-11
- NAIC, 2025 Cybersecurity Insurance Report (2024 data) (2025) checked 2026-11
- Lloyd's Market Bulletin Y5381, State backed cyber-attack exclusions (2022-08-16) checked 2026-11
- OFAC, Updated Advisory on Potential Sanctions Risks for Facilitating Ransomware Payments (2021-09-21) checked 2026-11
- New York DFS, Insurance Circular Letter No. 2 (2021), Cyber Insurance Risk Framework (2021-02-04) checked 2026-11
- GAO-22-104256, Cyber Insurance: Action Needed to Assess Potential Federal Response to Catastrophic Attacks (2022-06-21) checked 2026-11
- GAO-25-108748, Terrorism Risk Insurance Act: Considerations for Reauthorization (2025-09-23) checked 2026-11
- FTC, Cyber Insurance (Cybersecurity for Small Business) checked 2026-11
- SEC Release 33-11216, Cybersecurity Risk Management, Strategy, Governance, and Incident Disclosure (2023-07-26) checked 2026-11
- Merck & Co., Form 10-K for 2017 (2018-02-27) checked 2026-11
- MGM Resorts International, Form 8-K (2023-10-05) checked 2026-11
- Idaho Department of Administration, Risk Management Program, Insurance Coverages checked 2026-11
- NIST IR 8286 Rev. 1, Integrating Cybersecurity and Enterprise Risk Management (2025-12) checked 2026-11