Insurers hold some of the most sensitive data any business collects, including claims files with medical, financial and identity details, while depending on agents, brokers, third-party administrators and claims vendors who all need access to it. That combination has made the sector one of the most closely regulated for cybersecurity in the United States, through the NAIC Insurance Data Security Model Law (#668) adopted by many states and the New York DFS cybersecurity regulation (23 NYCRR Part 500), and in the EU through DORA. This guide sets out how to build an insurance cybersecurity programme that satisfies those rules: governance, risk assessment, third-party oversight, access and authentication, monitoring, incident notification and annual certification. It is for CISOs, compliance officers and security architects at carriers, MGAs, brokers and TPAs. Identity controls are covered in depth in the Insurance Sector Identity Security Guide and are summarised here, not repeated. Regulatory status checked on 27 October 2026.
Key takeaways
- Two US regimes set the bar. NAIC Model Law #668 (adopted in the fourth quarter of 2017, 21 states had adopted it by May 2024 according to the NAIC) and NYDFS Part 500, which is more prescriptive and applies to every DFS-licensed insurer.
- Notification clocks are short. Both regimes require notice to the regulator within 72 hours; Part 500 adds 24 hours for extortion payments with a written explanation within 30 days.
- Certification puts names on the line. NAIC domestic insurers certify by 15 February; Part 500 entities certify or acknowledge noncompliance by 15 April, signed by the highest-ranking executive and the CISO.
- Third parties are the sector's soft spot. Agents, brokers, TPAs, claims portals and SaaS integrations carry insurer data and credentials, and both regimes require due diligence and contractual controls.
- Universal MFA is now mandatory in New York. Since 1 November 2025, Part 500.12 requires MFA for any individual accessing any information system of a covered entity, with CISO-approved compensating controls as the only alternative.
Threats specific to insurers
| Threat | Why insurers are exposed | Priority controls |
|---|---|---|
| Credential theft against partner and agent portals | Thousands of external users (independent agents, brokers, adjusters, repair networks) log in to quoting and claims systems | Phishing-resistant MFA, device and session checks, rapid deprovisioning through partner feeds |
| Ransomware and extortion | Claims and policy systems are time-critical; sensitive data supports double extortion | Immutable backups, EDR, segmentation, tested recovery, extortion payment governance |
| Third-party and SaaS compromise | Claims processing, CRM, document management and print vendors hold nonpublic information | Vendor risk tiers, contract clauses, OAuth integration review, logging of vendor access |
| Claims and identity fraud | Account takeover of policyholder portals enables payment redirection | Step-up authentication for payment changes, behavioural fraud detection |
| Data aggregation in analytics and AI | Underwriting models and data lakes pool health, financial and telematics data | Data classification, minimisation and retention under Part 500.13(b), access reviews |
| Legacy core systems | Mainframe and decades-old policy administration platforms resist modern controls | Compensating controls, privileged access management, asset inventory with support end dates |
The Scania Financial Services incident shows the partner portal risk: in May 2025 an attacker logged in to an insurance claims application with a legitimate external user's credentials, which Scania believes were captured by infostealer malware, and downloaded claim documents. See the Scania Insurance Portal Breach.
The regulatory map for insurers
| Regime | Who it covers | Core cybersecurity duties |
|---|---|---|
| NAIC Insurance Data Security Model Law (#668), as adopted by each state | Licensees under the state's insurance law (insurers, producers and others); licensees with fewer than ten employees are exempt from Section 4 | Written information security programme based on risk assessment (Section 4), board oversight, third-party service provider oversight, investigation (Section 5), notification to the commissioner within 72 hours (Section 6), annual certification by domestic insurers by 15 February |
| NYDFS 23 NYCRR Part 500 (Second Amendment effective 1 November 2023) | Entities licensed or chartered under New York Banking, Insurance or Financial Services Law | Prescriptive programme: CISO, risk assessment, access privileges, MFA, asset inventory, monitoring and training, incident response and BCDR, 72-hour notice, annual certification |
| HIPAA Security Rule | Health plans and their business associates | Administrative, physical and technical safeguards for ePHI. NAIC #668 treats a licensee with a HIPAA-based programme that certifies compliance as compliant with Section 4 |
| EU DORA | Insurance and reinsurance undertakings in the EU, among other financial entities | ICT risk management, incident reporting, resilience testing, ICT third-party risk and register of information |
State adoption of #668 varies in wording and effective dates, so check each state's statute; the NAIC model leaves the effective date for states to fill in and gives one year to implement Section 4 and two years for third-party oversight (Section 4F). For deeper coverage see the NAIC Insurance Data Security Model Law Guide, the NYDFS Part 500 Guide, the HIPAA Security Rule Guide and the DORA Guide.
Building the programme: governance and risk assessment
- Board oversight: under #668 the board or a committee must require executive management to develop and run the programme and receive written reports at least annually. Part 500.4 requires the CISO to report to the senior governing body; use one report that satisfies both.
- Risk assessment as the anchor: #668 requires a programme based on a risk assessment that identifies reasonably foreseeable threats, including those from third-party service providers, and assesses safeguards at least annually. DFS issued further guidance on conducting and using Part 500 risk assessments on 10 September 2026; read it before your next cycle.
- Asset inventory: Part 500.13(a) requires a complete, accurate and documented inventory recording, where applicable, owner, location, classification or sensitivity, support expiration date and recovery time objective. Applicable since 1 November 2025.
- Data retention: Part 500.13(b) requires periodic disposal of nonpublic information no longer needed, unless law requires retention or targeted disposal is not reasonably feasible. Claims archives and model training data are the usual gaps.
- Size matters: Part 500 defines a Class A company as one with at least $20 million gross annual revenue in each of the last two fiscal years plus either over 2,000 employees averaged over the last two fiscal years or over $1 billion gross annual revenue in each of the last two fiscal years (affiliates count only where they share information systems, cybersecurity resources or programme elements). Class A companies carry extra duties for PAM, password blocking, EDR and centralised logging.
Access, authentication and identity
Access control is where examiners and attackers both focus. The key rules, with the identity detail in the Insurance Sector Identity Security Guide:
| Requirement | Source | Applies from |
|---|---|---|
| Least privilege, limited privileged accounts and use, annual access review, prompt termination, secure or disabled remote control protocols, written password policy | Part 500.7 | 1 May 2025 |
| Class A: privileged access management solution and monitoring of privileged activity; automated blocking of commonly used passwords (or CISO-approved compensating controls) | Part 500.7 | 1 May 2025 |
| MFA for any individual accessing any information system; CISO may approve equivalent or more secure compensating controls in writing, reviewed annually | Part 500.12 | 1 November 2025 |
| Access controls including MFA as a control that may be used for anyone accessing nonpublic information | NAIC #668 Section 4 | Per state adoption |
- Service accounts and integrations: Part 500.1 defines a privileged account to include service accounts that can perform security-relevant functions. Part 500.12's limited-exemption carve-out excludes service accounts that prevent interactive login, so do not try to apply MFA to them; instead vault, rotate and monitor them. See the NYDFS Part 500 Guide.
- Partners and agents: treat independent agents and TPAs as a separate identity population with federation or a dedicated partner identity provider, contractual MFA, and deprovisioning tied to appointment termination.
- OAuth and SaaS integrations: CRM and claims SaaS connectors hold long-lived tokens to insurer data. Inventory, scope and expire them. See the Gainsight Salesforce Breach for how long-lived integration tokens opened customers' CRM data.
Third-party service provider oversight
#668 Section 4F requires due diligence in selecting third-party service providers and requiring them to implement appropriate safeguards; Part 500.11 requires a written third-party service provider policy. A practical tiering model:
| Tier | Examples | Minimum controls |
|---|---|---|
| Critical | Claims TPAs, policy administration hosting, core SaaS, payment processors | Pre-contract assessment, MFA and encryption clauses, 72-hour or faster incident notice to you, right to audit, annual review, exit plan |
| High | Agencies and brokers with system access, document and print vendors | Security questionnaire, access via your identity controls, logging of access, notice clause |
| Standard | Vendors without nonpublic information | Baseline terms, periodic attestation |
Incident notice from vendors matters because Part 500.17 starts the 72-hour clock when the covered entity determines that an incident occurred at the entity, an affiliate or a third-party service provider. EU insurers also need the DORA register of information. See the DORA ICT Third-Party Risk and Register of Information Guide.
Monitoring, incident response and notification
- Monitoring and training: Part 500.14 requires monitoring of authorised user activity, controls against malicious code including web and e-mail filtering, and annual awareness training that covers social engineering. Class A companies must deploy EDR able to detect lateral movement and a centralised logging and alerting solution, unless the CISO approves equivalent controls in writing.
- Investigation: #668 Section 5 requires prompt investigation to determine the nature and scope of an event, identify the nonpublic information involved and restore security, with records kept for at least five years.
- Regulator notice: #668 Section 6 requires notice to the commissioner as promptly as possible and no later than 72 hours after determining an event occurred, where the state is the domiciliary or home state or where 250 or more of the state's consumers are involved and other conditions are met. Part 500.17 requires electronic notice to the superintendent within 72 hours.
- Extortion payments: Part 500.17 requires notice within 24 hours of payment and, within 30 days, a written explanation of why payment was necessary, alternatives considered and sanctions diligence, including OFAC rules. Build this into the ransomware playbook and the board's decision process. See the Ransomware Negotiation Guide.
- Consumer notice: follows state breach notification laws, and for health plans the HIPAA Breach Notification Rule.
How to implement: a phased approach
- Map obligations (month 1): list every state of licence and which have adopted #668, whether you are DFS-licensed, Class A status, HIPAA status and EU operations. Build one control set that meets the strictest requirement for each topic.
- Assess (months 1 to 3): run the risk assessment, build the Part 500.13 asset inventory and classify data stores, including claims archives and analytics platforms.
- Fix access first (months 2 to 6): universal MFA, PAM for privileged and service accounts, partner identity lifecycle, and common-password blocking. Pilot MFA with internal staff, then agents in waves, keeping a documented, time-limited fallback and CISO approval for any compensating control.
- Tier and contract vendors (months 3 to 9): re-paper critical vendors with notice, MFA and audit clauses; review OAuth integrations.
- Exercise (by month 9): a tabletop that runs both 72-hour clocks and the 24-hour extortion notice, with legal, claims operations and the board.
- Certify (annually): collect evidence through the year so the 15 February (NAIC domestic insurers) and 15 April (Part 500) filings rest on records, not recollection. See the NYDFS Part 500 Annual Certification Checklist.
Small versus large organisations: a small agency may be exempt from #668 Section 4 (fewer than ten employees) or qualify for Part 500's limited exemption, but must still apply MFA to remote access, cloud applications holding nonpublic information and privileged accounts under the Part 500 limited exemption. Large carriers should plan for Class A duties and run the programme through a GRC platform with control owners.
Measuring the programme
| Metric | Target |
|---|---|
| Human accounts covered by MFA (staff, agents, TPAs) | 100%, with documented compensating controls for exceptions |
| Privileged and service accounts vaulted and monitored | 100% for Class A; risk-based otherwise |
| Partner accounts removed within an agreed number of days of appointment termination | Above 98% |
| Critical vendors with current assessment and notice clause | 100% |
| Assets in inventory with owner, classification, support end date and RTO | Above 95% |
| Time from incident determination to regulator notice in exercises | Within 48 hours, leaving margin on the 72-hour clock |
Framework mapping
| Programme area | NAIC #668 | NYDFS Part 500 | NIST CSF 2.0 |
|---|---|---|---|
| Governance and board reporting | Section 4 | 500.4 | GV.OV-01, GV.RR-01 |
| Risk assessment | Section 4 | 500.9 | ID.RA-01 |
| Access and authentication | Section 4 | 500.7, 500.12 | PR.AA-01, PR.AA-03, PR.AA-05 |
| Asset inventory | Section 4 | 500.13 | ID.AM-01 |
| Third parties | Section 4F | 500.11 | GV.SC-05 |
| Monitoring | Section 4 | 500.14 | DE.CM-01, DE.CM-09 |
| Incident response and notice | Sections 4, 5, 6 | 500.16, 500.17 | RS.MA-01, RS.CO-02 |
Common failure patterns
- Forgetting the agents: internal staff get phishing-resistant MFA while thousands of agent and broker logins remain password-only.
- Vendor notice gaps: contracts give a TPA 30 days to report, which makes the insurer's 72-hour regulator deadline impossible.
- Service accounts treated as users: teams either try to force MFA on non-interactive accounts or exempt them entirely, rather than vaulting, rotating and monitoring them.
- Certification without evidence: a signed certification that cannot be backed by the five years of records both regimes require.
- Legacy systems missing from the inventory: mainframe and policy administration hosts with no support end date or RTO recorded.
- Long-lived integration tokens: CRM and claims SaaS connectors with broad scopes and no expiry.
Lessons from incidents
The Scania claims portal case combines the sector's two most common weaknesses: an external partner user whose credentials were stolen from an unmanaged device, and a portal run by an IT partner. Phishing-resistant MFA would have blunted replayed passwords, and session monitoring for bulk document download would have shortened detection. The Gainsight Salesforce breach shows the integration side: long-lived OAuth tokens held by a vendor opened customer CRM data, which is where many insurers keep policyholder and claims correspondence.
Practitioner checklist
- Map every state of licence to its version of #668, plus Part 500, HIPAA and DORA applicability.
- Confirm Class A status annually using the Part 500.1 thresholds.
- Run and document the annual risk assessment, including third-party threats.
- Enforce MFA for every individual, including agents and TPAs, and record CISO-approved compensating controls.
- Vault, rotate and monitor privileged and service accounts; block common passwords.
- Maintain the Part 500.13 asset inventory with owner, classification, support end date and RTO.
- Tier vendors and require incident notice fast enough to meet your 72-hour obligations.
- Embed the 72-hour, 24-hour and 30-day notification steps in the incident and ransomware playbooks.
- Keep evidence for at least five years and prepare certifications for 15 February and 15 April.
Standards and references
- NAIC: Insurance Data Security Model Law (#668)
- NAIC: Cybersecurity insurance topic page
- New York DFS: Cybersecurity Resource Center
- 23 NYCRR 500.1 Definitions
- 23 NYCRR 500.7 Access privileges and management
- 23 NYCRR 500.12 Multi-factor authentication
- 23 NYCRR 500.17 Notices to superintendent
- 23 NYCRR 500.22 Transitional periods
Related NHI Mgmt Group resources: Insurance Sector Identity Security Guide · NYDFS Part 500 Guide · NAIC Insurance Data Security Model Law Guide · Cyber Insurance Guide · DORA Guide. This guide summarises insurance sector cybersecurity regulation as at 27 October 2026 and is not legal advice.