Insurance by BogdanovaThe Encyclopedia of Insurance and Risk Science

RESEARCH DOSSIER

Data Subject Litigation and Class Action Cyber Liability

Abstract. Data subject litigation transforms compromises of confidentiality and unlawful data processing into a distinct liability exposure for organizations and their insurers. This monograph examines the transformation through individual recovery conditions, collective procedure, coverage, defense expenditure, reserving, and accumulation. Comparative documentary analysis centers on the European Union, United States federal courts and California, with narrower English and Ontario case comparisons. It uses legislation, selected judicial holdings, published insurance specimens, empirical research, and supervisory reporting. Conditional numerical examples show how legal eligibility, retentions, defense-related limit erosion, and common dependencies alter insurer payment distributions. The principal finding is that affected populations, compensable injuries, covered claims, and insurer payments are separate quantities linked through jurisdiction-specific legal and contractual conditions. Defense expenditure can create substantial insured losses without an award of damages. The available evidence establishes the mechanisms and selected historical observations but does not quantify current jurisdiction-specific litigation incidence or market distributions of defense costs. The contribution is an integrated analytical synthesis, rather than a new theory or empirical market estimate.

Keywords: Cyber insurance; data subject litigation; privacy liability; class actions; collective redress; GDPR; standing; defense costs; claims-made insurance; allocation; reserving; aggregation.

Introduction

The privacy liability insurance problem begins with the possibility that one information event will generate many legally distinct demands against one or more insureds over a prolonged period. A criminal may acquire copies of a repository without interrupting operations; a misconfigured system may expose data without an adversarial intrusion; a tracking application may process data unlawfully while functioning exactly as designed. All three pathways may produce civil allegations. The financial consequences depend on obligations imposed on organizations, available remedies, procedural routes, and the terms of insurance purchased. A claim for compensation is therefore not defined solely by the technical nature of the event.

Three questions guide the investigation. Which conditions make an affected person eligible to seek recovery? Through what procedural and economic mechanisms does collective litigation transform the scale and timing of loss? How do defense costs and allocation affect insured payments? A ruling on standing changes the legally relevant population without establishing coverage, while a policy clause restricting defense reimbursement may change insurer exposure without changing a claimant's rights. The study examines the connection between these domains.

Methodologically, the cases are purposefully selected contrasts rather than a representative sample. The policies are specimen documents, not executed contracts. Historical empirical studies have bounded sampling frames. The 2026 market report describes broad cyber claims rather than collective litigation alone. The research cutoff is November 18, 2026. Access ranged from extracted primary text to official indexed descriptions; the evidence ledger records these differences. No confidential insurer data, original court microdata replication, or comprehensive update of every jurisdiction's law is claimed. All numerical scenarios are explicitly illustrative.

Conceptual and Historical Foundations

The legal and financial objects

A data subject is a person to whom personal data relate. The EU General Data Protection Regulation (European Parliament and Council, 2016) supplies the relevant distinctions between personal data, data subject, controller, and processor. The relation between an affected person and the insured may be indirect. A retailer can determine the purposes of processing, an outsourced technology company can process records on its instructions, and an employee can be both a data subject and potential claimant. Commercial databases also contain proprietary information outside personal-data law. The present inquiry concerns civil claims involving identifiable people and connected confidentiality allegations.

The material event is a change in the state or use of information. A breach may involve unauthorized access or disclosure, while unlawful processing can occur without outside interference. A legal wrong depends on an applicable obligation and conduct satisfying the relevant liability conditions. An actionable injury depends on the cause of action and forum. A claim under a policy is defined by contract and may arise upon receipt of a demand before responsibility has been established. A covered loss is the designated expenditure surviving applicable exclusions and financial terms.

These levels have different denominators. A forensic report may count compromised files, identifiers, accounts or records. One person can have numerous records; one household account may contain information about several people. A proceeding may represent thousands of individuals, while multiple proceedings can include the same persons. Insurers can open separate files for different insureds or policies associated with one event. The number of exposed files cannot be equated with the number of injured persons, legal actions, insurance claims, or payments.

Historical US litigation illustrates the importance of distinguishing injury and procedure. The historical docket study by Romanosky, Hoffman, and Acquisti (2014) examined federal breach litigation during 2000–2010. That observation period predates contemporary GDPR litigation and California's current private statutory remedy. Its value is to support investigation of litigation selection and outcomes, not to provide present-day claim probabilities.

Security evidence and legal responsibility

Forensic analysis should identify the information asset, retention history, access privileges, transfer mechanisms, and time of discovery. Evidence that an attacker reached a database does not necessarily establish which records were extracted. Missing logs create an evidentiary limitation. A report of later identity fraud may be consistent with compromise yet also have alternative causes. The event investigation should therefore maintain a distinction between directly observed exposure, inferred exposure, and subsequent harm.

The 2026 ENISA Threat Landscape identifies a 2025 observation window (ENISA, 2026). Threat observations follow their own collection criteria. An agency-observed event, a breach requiring notification, a lawsuit, and an insurer-reported claim are distinct observations. Linking these datasets requires shared event identifiers and timing; summation without linkage could count the same occurrence repeatedly or misattribute a demand.

Notification and mitigation introduce further distinctions. A person can receive a breach notice without establishing present harm. Forensic, notification, legal advice, and monitoring costs may arise before any opposing demand. Notification can inform potential litigants, whereas effective mitigation may reduce some later damage. These mechanisms may operate simultaneously. Comparing organizations that notify or offer monitoring with organizations that do not reveals associations unless the assignment process and preexisting risk differences are adequately controlled.

Historical purpose of collective procedure

Collective proceedings allow individuals to share the cost of asserting common claims. Evidence about system design, corporate conduct, acquisition pathways and the general control environment may be common across claimants. Actual identity misuse, individualized financial loss, residency, consent and reliance may differ. Aggregation lowers some costs of proof without necessarily making the underlying rights uniform.

The insurance implications are two-sided. Collective filing can create defense expenditure before certification, while coordinated proceedings can eliminate duplicate discovery and provide closure for a defined population. Broader classes, multiple forums and extensive remedies may increase expense. Shared expertise and common proof can reduce redundant work. Claim frequency, defense severity and indemnity severity must therefore be examined separately.

Primary Documents and the Scope of Data Subject Standing

European Union: individual recovery and collective representation

The GDPR liability provisions (European Parliament and Council, 2016) connect organizational data protection duties with rights and remedies. Article 82 concerns compensation for material and nonmaterial damage; processor responsibility has a particular scope. Articles 79 and 80 address judicial remedies and representation. National procedure, applicable law, processing facts and time of conduct remain relevant to specific cases.

In Österreichische Post (CJEU, 2023), the Court identified infringement, damage and causal connection as necessary compensation conditions. Nonmaterial harm does not need to exceed a separate minimum level of seriousness. That holding distinguishes existence of damage from a seriousness threshold. It does not supply a uniform monetary amount. A financial model using Article 82 must maintain the distinction between people affected by disputed processing and those who establish compensable damage.

The third-party intrusion in VB v Natsionalna agentsia za prihodite (CJEU, 2023) concerned security appropriateness, responsibility and fear of potential misuse. The fact of attack alone does not settle whether safeguards were appropriate. Fear of misuse may constitute nonmaterial damage under the relevant conditions. Consequently, both the actual security environment and the individual circumstances of harm matter. Estimation from a record count alone cannot establish a compensable population or valuation.

The Meta Platforms Ireland judgment (CJEU, 2022) concerns a consumer association acting without individual mandates in the setting of national consumer law. The judgment addresses the availability of representative action, not automatic class-wide damages. The Representative Actions Directive (European Parliament and Council, 2020) establishes a framework for collective consumer redress through qualified entities. National implementation, covered rights, participation and remedy must be checked for a particular forum.

These legal mechanisms can increase the number of claims that are practically enforceable without an equivalent change in the number of records exposed. The costs of common issues may arise where individual damages remain disputed; injunctions can require remediation rather than compensation. A policy defining a claim broadly may therefore transfer defense expenses even when monetary recovery is limited.

United States: standing and substantive private rights

In TransUnion LLC v. Ramirez (U.S. Supreme Court, 2021), a Fair Credit Reporting Act dispute, the Court distinguished members whose misleading credit reports were disclosed to third parties from members whose information remained internal. Of 8,185 class members, 1,853 had reports disseminated during the relevant time. The Court found standing for that population on the reasonable-procedures damages claim and reached a different result concerning the other asserted claims. This is a standing precedent rather than an empirical cyber breach example.

Federal Article III standing and the merits are distinguishable. A retrospective damages claimant must establish concrete injury in the context of the asserted claim. The standing assessment for prospective relief differs. State courts may operate under different requirements. Insurers should consequently distinguish the claims alleged, those legally eligible for adjudication, and those that survive substantive defenses; the first set can generate expenses before the others are determined.

California's Civil Code § 1798.150 (California Privacy Protection Agency, 2026) provides a private remedy for specified compromised personal information when qualifying unauthorized access and exfiltration, theft, or disclosure results from a business's failure to maintain reasonable security. The provision concerns specified security breaches rather than every privacy violation. It includes actual and statutory damages, injunctive or declaratory relief, and a notice-and-cure mechanism for statutory claims. Post-breach adoption of reasonable safeguards alone does not constitute cure of that breach.

The indexed monetary thresholds (California Privacy Protection Agency, 2024) establish that from January 1, 2025, the statutory range is $107–$799 per consumer per incident, or actual damages if greater. The printed base amounts of $100–$750 must therefore be distinguished from the currently adjusted range. The applicable date, eligibility and judicial assessment remain decisive. The statutory range does not identify what a particular settlement will pay.

A scheduled range tied to eligible consumers creates a different arithmetic exposure from compensation assessed through individualized proof of harm. Both approaches require legal selection of the relevant population. Neither the California remedy nor Article 82 can be generalized to all neighboring jurisdictions.

England and Ontario: selected comparative controls

In Lloyd v Google (UK Supreme Court, 2021), uniform damages were sought under the historical Data Protection Act 1998 through a representative procedure. Individual financial damage or distress and the limitations of the selected procedure were central to the case. The dispute illustrates the distinction between a common issue of wrongful processing and uniform monetary entitlement. It concerns historical law and should not be treated as a comprehensive account of current English privacy remedies.

The official docket summary of Owsianik v Equifax (Supreme Court of Canada, Office of the Registrar, 2023) records that intrusion-upon-seclusion common issues against the database custodian were struck after an intrusion by third parties. The Ontario appeal was dismissed and the subsequent leave application refused. The docket account supports a bounded distinction between allegations of the custodian's security failings and commission of the intrusion itself. It does not provide the underlying court's complete legal reasoning, and the refusal of leave is not an affirmance of every proposition in that reasoning. Other causes of action remain separate.

The comparison supports a practical conclusion: identical technical evidence may enter different legal tests and lead to different eligible populations and remedies. An international insurer must preserve forum and cause-of-action identifiers in exposure analysis. A single global standing factor conceals the very legal variation that determines the risk.

Mechanisms and Evidence for Collective Litigation

Certification and overlapping proceedings

US federal Rule 23 distinguishes the common prerequisites for class treatment from the required category of action. Damages classes under Rule 23(b)(3) generally require predominance of common questions and superiority of class treatment, with notice and exclusion opportunities. Settlement of a certified or proposed settlement class requires judicial approval under Rule 23(e). Certification is a procedural decision and does not itself establish ultimate liability.

Several actions may relate to one breach. Coordination can reduce duplicated discovery without ensuring that all claimants and defendants are included in one final release. An individual outside the class definition, a person opting out, or a claimant with a different cause may maintain separate rights. A claims inventory should associate each proceeding with its proposed and actual population, remedy, defendants and stage. The total number of alleged class members cannot simply be added when overlapping actions exist.

The economics of aggregation are conditional. Common evidence can reduce cost per represented person, while reduced filing costs may expand the represented population. Complex certification disputes and discovery create expenses that separate claims might not generate. A credible comparison of collective and individual procedures would need to include potential claimants who would never file independently, rather than assuming all would litigate in the absence of aggregation.

Empirical selection and causal limitations

Romanosky, Hoffman, and Acquisti (2014) analyzed filing and settlement patterns in more than 230 federal breach lawsuits. Their results linked financial harm with higher odds of litigation, free credit monitoring with lower odds, and financial-loss allegations or certification with more frequent settlement. The historical associations do not prove that offering monitoring universally prevents litigation or that certification itself causes a fixed change in settlement probability.

Organizations differ in resources, baseline safeguards and exposure. Severe incidents may prompt both more generous mitigation and more litigation. Well-managed organizations may both offer monitoring and have lower residual risk. The resulting bias can operate in either direction. Similarly, a legally stronger case may both be certified and settle because of latent case quality. A causal study would require a defined cohort, a treatment assignment model, measurement of confounders and a clear temporal ordering.

Romanosky, Ablon, Kuehn, and Jones (2019) conducted a qualitative policy filing analysis examining coverage, exclusions, security applications, and rating schedules. The historical filing sample establishes the feasibility of investigating policy language systematically; it is not a current market census. Its methodology supports the use of identifiable contract editions and explicit variation in terms when conducting cyber liability research.

Settlement administration and exposure

The FTC's Equifax settlement information (2024 update) records about 147 million affected individuals and up to $425 million in assistance. The figures relate to the incident and overall consumer relief, not the insurance indemnity or defense expenses of a particular insurer. The administration of different types of claims extends beyond the initial announcement.

Settlement amounts may include reimbursement, services, administrative expense and other components. Headline figures can represent commitments or ceilings rather than actual distributions. Participation is incomplete in some claims programs and unused funds may be handled under specific court-approved terms. The affected population, eligible class, submitted claims and paid claims therefore require separate records. An announced aggregate divided by all affected persons is not evidence of the average amount received by actual claimants.

Contractual translation

The AmTrustCyber 2021 specimen policy distinguishes discovery-based direct-loss protection from claims-made third-party coverage and places defense expenses within the aggregate limit. It includes notice, consent and circumstance-reporting provisions. Thus the incident, discovery, demand, insurer notice and payment dates can be different. A reporting extension may affect eligibility without increasing aggregate capacity.

The AIG CyberEdge 2013 specimen wording defines claims to include specified written demands, lawsuits and certain requests for mediation or tolling; it also addresses hosted resources and insured consent obligations. The specimen demonstrates why a claim register may need to begin before a complaint is filed. These identified editions are examples, not universal market terms. Declarations, endorsements and applicable law can change their operation.

An insured can hold cyber, technology E&O, management liability and other insurance. A lawsuit may allege insecure processing, misleading statements and contractual nonperformance. Coverage assessment requires associating each allegation with the insured entity, insuring clause, definition, exclusions, reporting period and retention. Separate policies can interact through allocation and other-insurance conditions. A technical event label cannot determine the contract response by itself.

Comparative Measurement of Defense Costs

Counting exposure and expenditure

Let N denote unique affected persons, E the persons satisfying the selected legal eligibility conditions, and A those within the relevant represented or participating population. In an opt-out class, inclusion does not require each member to submit a claim; distribution procedures may impose separate requirements. The sets and their dates must be defined before estimating recoverable amounts.

Let D represent eligible defense expense, M covered damages or settlement, U uncovered costs, L the remaining aggregate before this event, and R a single joint retention. Assuming no additional sublimits, prior erosion or other coverage, net insurer payment is min(L, max(0, D + M − R)). The insured retains D + M + U minus insurer payment. This is a conditional accounting identity under an assumed contract, not an interpretation of every cyber policy.

Where M equals zero, defense may still be substantial. Where defense is outside the limit or has a separate cap, the formula changes. Eligible expenses must be distinguished from total billed expenses; allocation precedes application of the hypothetical financial terms. Paid defense, incurred defense, outstanding defense and damages reserves can develop at different rates.

Synthetic defense-erosion scenario

Assume a $5 million aggregate, one $250,000 retention, $1.4 million in covered defense and $5.2 million in covered settlement. Both categories share the aggregate. Covered total is $6.6 million; after retention, $6.35 million remains, of which the insurer pays $5 million. The insured retains $1.6 million including retention and excess of limit. Allocating the retention to defense first produces $1.15 million in insurer-paid defense, leaving $3.85 million of aggregate capacity for damages.

Assume instead that defense is payable outside a $5 million damages limit without a separate defense cap, with the same retention ordering. Insurer payment becomes $1.15 million in defense plus $5 million in damages, totaling $6.15 million. The insured retains $450,000. The difference is the amount of insurer-funded defense that otherwise erodes the indemnity limit. If the suit is dismissed after $1.4 million of eligible defense expense, the original design still produces $1.15 million of insurer payments.

These examples show why defense expenses cannot be projected exclusively as a percentage of awarded damages. A jointly modeled distribution may recognize dependence between complexity, procedural duration and damages, while preserving the possibility of high defense expenses and zero indemnity. A fixed ratio would miss this outcome.

Synthetic statutory-damages scenario

Assume 100,000 unique affected people and 40,000 who satisfy all relevant California statutory conditions and are included in a hypothetical compensable group for one incident. Using the 2025 adjustment table, the conditional arithmetic range is 40,000 × $107 = $4.28 million to 40,000 × $799 = $31.96 million. If only 10,000 qualify, the corresponding figures become $1.07 million and $7.99 million.

These are conditional amounts before defenses, actual damages exceeding the statutory range, litigation costs, settlement and insurance terms. They are neither predicted awards nor unconditional limits of liability. In a hypothetical compensatory regime, 10,000 persons establishing harm and causation with an assumed mean payment of $200 would produce $2 million. The $200 is an illustrative parameter, not a jurisdictional statistic. The example demonstrates why eligibility and valuation should be measured under the appropriate remedy before applying coverage.

Claims development and market denominators

A litigation-specific loss triangle can group by report year and development lag and separately accumulate defense and damages. For claims-made insurance, notice dates help identify reporting cohorts; occurrence and discovery dates help evaluate latent exposure. Dismissed cases should remain in defense distributions, and open severe cases must not disappear from sample calculations. A recent cohort's low paid severity may reflect incomplete development rather than lower ultimate cost.

The NAIC (2026) cyber insurance market report records 5,843,175 domestic policies in force in 2025, compared with 4,368,614 in 2024, and 56,323 reported claims in 2025. Dividing the latter by the 2025 stock yields about 0.964 claims per 100 policies. This is descriptive arithmetic across broad categories, not a probability of an incident, class-action incidence, or an earned-exposure frequency. The denominator includes differing coverage types and the numerator is not litigation-specific. An incident may affect multiple policies.

A proper dataset would identify legal stage, overlapping proceedings, insured defendants, hours and approved legal fees, expert and forensic costs, covered and uncovered allocations, incurred and paid expense, and recoveries. Comparison across countries requires consistent currencies, valuation dates, court costs and treatment of legal fee shifting. A broad market ratio and a headline settlement cannot substitute for such data. The absence of disaggregated, longitudinal defense-cost data limits numerical conclusions in this study.

Integrated Scientific Synthesis and Research Agenda

From information exposure to insurer payment

The consistent analytical result is a sequence of transformations. A technical event changes the set of persons potentially affected. Substantive law determines which facts may satisfy a claim. Procedure determines which disputes can be adjudicated together and which persons are included. Coverage determines whether specified expense is transferable. Financial limitations and recoveries determine net payment. Each transformation uses a different observation unit and decision date.

The relationship is many-to-many: persons to incidents, incidents to defendants, proceedings to persons, insureds to policies, and policies to payments. An exposure register should preserve these links, rather than add overlapping class definitions. Directly attributable defense tasks should be separated from shared tasks. Gross defense, covered expense, insurer payment and uncovered remainder should reconcile. These are applications of ordinary legal and insurance accounting distinctions to the specific problem of collective cyber liability.

The selected jurisdictions establish why a uniform eligibility parameter is inadequate. European damages analysis considers infringement, damage and causation. California's specified breach remedy includes inflation-adjusted statutory amounts. US federal standing can narrow the damages population, and the selected representative-action decision illustrates individualized proof problems. The two contract specimens demonstrate conditions under which a demand generates defense expenditure before adjudication. These observations do not establish a universal event-to-claim conversion rate.

Research agenda and measurement design

A prospective study should observe a defined cohort of insured and self-insured organizations across named jurisdictions, including incidents producing no litigation. Each event should carry information category, unique persons affected, exposure uncertainty, corporate scale, procedural jurisdiction, causes of action, remedy, and development status. Coverage data should contain form editions, endorsements, retentions, limits, and defense treatment. Outcome data should record actual expense and payments at common valuation dates, keeping open and closed cases separate.

An analysis of mitigation must predefine whether the desired effect concerns filing, settlement, claimant harm, or insured payment. Comparing recipients and nonrecipients of monitoring requires explicit assumptions about treatment assignment, confounding and follow-up. Legislative changes may create opportunities for quasi-experimental analysis if simultaneous policy changes and prior trends can be addressed. Neither design automatically identifies causal effects.

An analysis of aggregate procedure should compare coordinated and noncoordinated cases after accounting for class size, complexity, selection, jurisdiction and duration. Total legal expense and expenditure per eligible claimant can move in different directions. Distribution of relief and time to payment should also be examined. A study confined to insurer expenses would omit important social consequences.

Research records should preserve privacy through data minimization, pseudonymous event and claimant keys, and appropriate access restrictions. Any public empirical result should identify the cohort, observation window, missingness, policy categories, jurisdiction definitions and valuation basis. A future claims dataset could test the mechanisms presented here rather than treating synthetic scenarios as observations.

Evidence boundaries and original contribution

The documentary comparison establishes mechanisms and selected historical facts, not a comprehensive current legal survey. Official EU texts and several judgments were accessible only through indexed passages or abbreviated retrieval. Rule 23 and the complete English judgment were not directly retrieved. The Ontario example rests on the court's docket summary. The specimen forms are historical and do not establish contemporary prevalence. ENISA supports only the stated collection period; the NAIC report does not separate class litigation. No observed distribution of defense costs or causal effect of collective procedure is claimed.

The original synthesis relates jurisdiction-specific legal eligibility and collective procedure to insurance exposure denominators, separate defense-cost development, overlapping matters, contractual limit erosion and dependence. The conditional calculations demonstrate these consequences under explicit assumptions. No original named doctrine, priority claim or novel empirical parameter is asserted.

Conclusion

Data subject litigation is a distinctive liability exposure because one information event can generate many potential demands while defense costs arise before entitlement to compensation is determined. Eligibility is specific to the legal system, cause, forum and remedy. The affected population, legal claimant, covered claim and insurance payment remain separate units of analysis.

Collective action can expand enforceability and reduce duplicate proof. Its financial consequences depend on the grouping of claims, individual injury requirements, procedural development and settlement administration. Defense expenditures have their own drivers, and eroding limits can materially reduce available indemnity. Insurance assessment must therefore identify policy editions, preserve exposure and claim relationships, and separate gross legal costs from covered expense and net payment.

The selected evidence supports an integrated explanation of the mechanisms and their historical context. Stronger empirical estimates require linked event and litigation cohorts, contract data, actual defense budgets, distributions of relief and loss development. Transparent legal scope, evidence quality and explicit contractual assumptions make comparative insurance science possible without mistaking a technical record count for a financial loss estimate.

References

  • AIG. (2013 form edition). CyberEdge Security and Privacy Liability Insurance, General Terms 101013 (12/13) and Security and Privacy Section 101024 (12/13). Specimen. Source.

  • California Privacy Protection Agency. (2024, December 17). Updated Monetary Thresholds in CCPA (effective January 1, 2025). Source.

  • California Privacy Protection Agency. (2026). California Consumer Privacy Act of 2018, statutory text effective January 1, 2026, § 1798.150. Source.

  • Court of Justice of the European Union. (2022, April 28). Meta Platforms Ireland Ltd v Bundesverband der Verbraucherzentralen und Verbraucherverbände, C-319/20, ECLI:EU:C:2022:322. Source.

  • Court of Justice of the European Union. (2023, December 14). VB v Natsionalna agentsia za prihodite, C-340/21, ECLI:EU:C:2023:986. Source.

  • Court of Justice of the European Union. (2023, May 4). UI v Österreichische Post AG, C-300/21, ECLI:EU:C:2023:370. Source.

  • European Parliament and Council. (2016). Regulation (EU) 2016/679 (General Data Protection Regulation). Official Journal L 119, 1–88. Source.

  • European Parliament and Council. (2020). Directive (EU) 2020/1828 on representative actions for the protection of the collective interests of consumers. Official Journal L 409, 1–27. Source.

  • European Union Agency for Cybersecurity. (2026, September 22). ENISA Threat Landscape 2026. Source.

  • Federal Trade Commission. (2024, November update). Equifax Data Breach Settlement. Source.

  • National Association of Insurance Commissioners. (2026). Report on the Cybersecurity Insurance Market, calendar year 2025. Section 1, Figures 3–5. Source.

  • Romanosky, S., Ablon, L., Kuehn, A., & Jones, T. (2019). Content analysis of cyber insurance policies: How do carriers price cyber risk? Journal of Cybersecurity, 5(1), tyz002. DOI: 10.1093/cybsec/tyz002. Source.

  • Romanosky, S., Hoffman, D., & Acquisti, A. (2014). Empirical analysis of data breach litigation. Journal of Empirical Legal Studies, 11(1), 74–104. DOI: 10.1111/jels.12035. Source.

  • Supreme Court of Canada, Office of the Registrar. (2023). Alina Owsianik v Equifax Canada Co., docket 40577; underlying decision 2022 ONCA 813. Source.

  • Supreme Court of the United States. (2021, June 25). TransUnion LLC v. Ramirez, 594 U.S. 413, No. 20-297. Source.

  • Technology Insurance Company, Inc. (2021 form edition). AmTrustCyber sample policy, CYDEC 0321 and CY990001 0321. Victor Insurance specimen. Source.

  • United Kingdom Supreme Court. (2021, November 10). Lloyd v Google LLC, 2021 UKSC 50. Judgment reproduced by BAILII. Source.

  • United States Federal Rules of Civil Procedure. Rule 23, Class Actions. Cornell Law School, Legal Information Institute (online reproduction). Source.