Why insurance cybersecurity matters in 2026
In my experience working with US P&C carriers between $500M and $5B GWP, the cybersecurity question has moved from 'should we spend more' to 'where does the next dollar go.' The numbers behind that shift are concrete:
- A data breach in financial services, the industry group closest to insurance in IBM's 2026 Cost of a Data Breach Report, now costs $6.29 million on average, up from $5.56 million a year earlier. The US average across all industries reached $11.5 million.
- Third parties were involved in 48% of breaches in Verizon's 2026 Data Breach Investigations Report, a 60% jump in one year. In the Financial and Insurance sector the figure was 34%.
- The 2024 ransomware attack on Change Healthcare, a health care claims and payment processor owned by UnitedHealth Group, affected approximately 192.7 million people, according to the HHS Office for Civil Rights.
- New York collected $11.3 million from GEICO and Travelers in November 2024 (NY Attorney General and DFS) and more than $19 million from eight more auto insurers in October 2025 (Insurance Journal), all tied to data stolen through online quoting tools.
Carriers hold exactly the data attackers want: Social Security and driver's license numbers, bank details, medical records in bodily injury and workers' compensation files, and claim histories. What stands out to me in the New York cases is how basic the failures were. These were not nation-state operations. They were missing multi-factor authentication and driver's license numbers exposed through quoting tools.
That is why cybersecurity in insurance software is now a board-level platform decision, not an IT line item. If security sits inside a broader core modernization program at your carrier, our mid-tier carrier modernization framework shows where security decisions fit next to platform choices.
What cybersecurity in insurance software covers
Cybersecurity in insurance software is the set of technical and organizational controls built into policy administration, billing, claims and distribution systems to protect policyholder data, claim records and financial transactions from unauthorized access, theft, ransomware, fraud and disruption. In 2026 the baseline includes encryption at rest and in transit, multi-factor authentication for every user, least-privilege access, a documented inventory of systems and data, continuous monitoring, third-party oversight and a tested incident response plan that satisfies NY DFS Part 500 and state laws based on the NAIC Insurance Data Security Model Law.
The defining change is where those controls live. Security is designed into the platform from the start: identity checked on every API call, encryption in every data store, an audit trail on every transaction and monitoring that runs continuously. For a feature-by-feature checklist, see key security features in insurance software.
The 2025-2026 threat landscape: ransomware, social engineering, third parties
Three threat categories shape decisions about cybersecurity in insurance software for US carriers right now.
Ransomware
Ransomware appeared in 48% of breaches in Verizon's 2026 DBIR, up from 44% a year earlier, although 69% of victims in that dataset refused to pay. Chainalysis tracked about $820 million in ransom payments in 2025 while claimed attacks grew 50% and the payment rate fell to a record low of 28%, as The Record reported. Fewer victims pay, but more get hit.
Change Healthcare is the case insurance executives should study. UnitedHealth's CEO told the Senate Finance Committee that attackers got in through a server without multi-factor authentication, that the attack also locked up backup systems, and that the company paid a $22 million ransom (The Record). In my experience, the lesson for P&C carriers is less about the ransom and more about legacy: the breached environment came from an acquired company running older technology, and that is exactly where most mid-tier carriers carry their own unmonitored risk.
Social engineering and business email compromise
The human element was present in 65% of Financial and Insurance breaches in the 2026 DBIR, and pretexting, often by phone, has become a more common way into ransomware and extortion attacks. Business email compromise produced $3.05 billion in reported losses across 24,768 complaints in the FBI's 2025 Internet Crime Report.
A common pattern in insurance is a request to change bank details for a vendor, repair shop or claimant, timed just before a large payment. Voice cloning and AI-written emails make those requests harder to spot. The controls that work are procedural as much as technical: out-of-band verification for any change to payment details, dual approval above a threshold, and payment workflows in the claims and billing systems that an email cannot override.
Third-party and supply chain risk
Carriers connect to brokers, MGAs, reinsurers, TPAs, prefill data providers and cloud platforms, and every connection widens the attack surface. The New York auto insurer cases show what this looks like in practice. According to the New York Attorney General, attackers used compromised agent credentials to log into a Travelers agent portal that had no multi-factor authentication, generated reports with full driver's license numbers in plain text, and went undetected for more than seven months until a third-party prefill data provider raised the alarm.
In my experience, agent and broker portals are the most under-protected surface in a mid-tier carrier, because they are built for convenience and shared with thousands of external users. Part 500 already requires written third-party service provider policies, so the gap is usually execution, not policy. We cover the portal side in security and compliance in insurance agent portals.
Regulatory requirements: NY DFS Part 500, NAIC Model Law #668, HIPAA
The cybersecurity rules for US carriers tightened between 2023 and 2025, and enforcement followed:
- NY DFS 23 NYCRR Part 500 applies to every insurer licensed in New York. The Second Amendment, adopted in November 2023, phased in through November 1, 2025, when multi-factor authentication for any individual accessing any information system and a documented asset inventory became mandatory. Part 500 also requires a CISO, board-level oversight, encryption of nonpublic information, third-party service provider policies, notice to DFS within 72 hours of a cybersecurity event and an annual compliance certification. Both New York settlements described above were brought under it.
- NAIC Insurance Data Security Model Law (#668) has been adopted in 28 jurisdictions, according to the NAIC's March 2026 state implementation report. It requires a written information security program based on a risk assessment, oversight of third-party service providers, investigation of cybersecurity events and notice to the state insurance commissioner (within 72 hours in the model text).
- HIPAA Security Rule applies to health plans and their business associates (HHS summary of the Security Rule). Most P&C carriers are not HIPAA covered entities, but medical records in bodily injury and workers' compensation files still call for equivalent safeguards under state law.
- Other frameworks depend on your footprint: GLBA privacy and safeguards rules, CCPA/CPRA for California residents, PCI DSS for card payments, SOX for publicly listed carriers, and GDPR or UK GDPR (with 72-hour breach notification) for EU and UK data.
- AI governance: if AI influences underwriting, claims or fraud decisions, the NAIC Model Bulletin on the use of AI systems by insurers, adopted in 25 jurisdictions as of March 2026, adds documentation, testing and third-party AI vendor oversight on top of security controls.
In my experience, Part 500 works as the de facto national benchmark: a carrier that can pass a DFS examination rarely struggles in a Model Law state. For the state-by-state picture, see regulatory compliance in US insurance, and for a broader checklist that covers ISO, AAIS and state filings, see compliance-ready insurance software.
AI and machine learning for cybersecurity defense
AI now sits on both sides of the fight. IBM found that one in four malicious breaches in its 2026 study were AI-enabled, a 56% increase, mostly deepfake impersonation and AI-enabled malware, and that these breaches cost about $6 million on average (IBM press release). On the defense side, organizations using security AI and automation cut breach costs by almost $2 million. What AI does well in insurance security:
- Anomaly detection - flags deviations in logins, API calls and database queries, such as an agent account pulling thousands of quotes overnight.
- Phishing and impersonation detection - spots AI-written emails, spoofed domains and voice deepfakes aimed at payment staff.
- Data flow monitoring - tracks where policyholder data moves and flags unexpected exports.
- Automated containment - locks accounts, isolates hosts or revokes API keys within seconds, with human review afterwards.
- Insider threat detection - identifies behavior that suggests compromised credentials or misuse of access.
Skills are the real constraint. In ISC2's 2025 Cybersecurity Workforce Study, 88% of respondents said their organization had suffered at least one significant cybersecurity consequence in the past year because of a skills shortage. My take: I'd recommend applying AI first to identity and payment flows, where the IBM and FBI numbers show the money is actually lost, before buying a broad AI security platform. For the wider picture of where AI fits in carrier operations, see AI for insurance.
Zero Trust architecture and identity management
Zero Trust, as defined in NIST SP 800-207, removes implicit trust based on network location or asset ownership: every user, device and service is authenticated and authorized before it opens a session with a resource. The components that matter in insurance:
- Identity-based access - every API call carries a verified identity and is checked against current policy.
- Multi-factor authentication - for employees, agents, brokers and administrators, with phishing-resistant methods for privileged accounts.
- Least privilege - users and services get only the data and functions their role needs.
- Segmentation - a compromise in one zone, such as an agent portal, cannot spread freely to policy or claims data.
- Continuous verification - sessions are re-checked rather than trusted for their full length after login.
- Encryption everywhere - at rest and in transit by default, plus field-level encryption or tokenization for Social Security, driver's license and bank account numbers.
Part 500 does not use the words Zero Trust, but its requirements point the same way: universal MFA, limits on privileged access, an asset inventory and monitoring of user activity. In my experience, carriers that present their Part 500 program to the board as a Zero Trust roadmap get budget approved faster, because the board sees one plan instead of a list of findings. Modular design also limits how far an intrusion can travel; see security implications of modular architecture in insurance systems.
Cybersecurity during data migration
Data migration is the moment when cybersecurity in insurance software is tested hardest. Cloud moves, legacy PAS replacement, M&A integration and book transfers all put data in motion. Typical risks include:
- Data exposed in transit or in staging databases with weaker controls than production.
- Real Social Security and bank account numbers copied into test environments.
- Migration tooling logs that capture full policyholder records.
- Temporary bypasses of security controls to make old and new systems talk to each other.
- Contractors and consultants with elevated access that nobody revokes after cutover.
- Integrity gaps where migrated data no longer matches the source system.
When Decerto migrated an acquired insurer's data into Generali Group Poland's systems, the cutover itself took a single weekend, but only after field-by-field analysis and multiple rounds of trial migrations (Generali Group Poland case study). In my experience, trial runs are also where security gaps surface, so I'd require masked data in every non-production run and a named owner for every temporary access right.
Purpose-built tooling, including Decerto's Data Migrator, combines data mapping, automated validation and migration result reporting, which gives security and audit teams a record of every transfer. For the full checklist, see data security during data migration in insurance.
How Decerto builds security into insurance platforms
Decerto's approach to cybersecurity in insurance software is to build security into the insurance platform rather than bolt it on as a separate layer. In practice that means:
- Encryption of policyholder and claim data in transit and at rest.
- Secure login with MFA, role-based access control and least privilege.
- An audit trail on every transaction - policy change, claim update, rule change, AI-supported decision - showing who, what, when and why, ready for state DOI examinations.
- Authenticated, monitored APIs for brokers, data providers and other integration partners.
- Regular security audits and documented incident response procedures aligned with the carrier's Part 500 and Model Law obligations.
- Data Migrator for legacy-to-modern transfers with validation and reconciliation.
The project I point to most often is Warta (HDI/Talanx Group). Its eAgent platform, built by Decerto, serves more than 40,000 agents, makes over 1 million daily requests to external data sources and has run for more than 30 consecutive months without a service availability issue. Every one of those external calls is an integration point that has to be authenticated, logged and monitored. The same principles apply to core systems; see compliance and security in insurance policy administration systems.
Honest disclosure: Decerto is not a specialized cybersecurity vendor. Endpoint protection, SIEM, network security and threat intelligence belong to the dedicated security tools your CISO already runs. Our job is the security architecture inside the insurance platform and its integration with that stack. For NY DFS-regulated entities, health-adjacent lines with HIPAA exposure and carriers holding EU or UK data, we work directly with the carrier's CISO team.
FAQ
Cybersecurity in insurance software covers thecontrols built into policy, billing, claims and distribution systems:encryption at rest and in transit, multi-factor authentication for every user,least-privilege access, an asset inventory, continuous monitoring, third-partyoversight and a tested incident response plan. Together they protectpolicyholder data and support NY DFS Part 500 and NAIC Model Law compliance.
IBM's 2026 Cost of a Data Breach Report putsthe average breach in financial services, the industry group closest toinsurance, at $6.29 million, and the US average across all industries at $11.5million. Regulatory penalties come on top: New York collected more than $30million from ten auto insurers in 2024 and 2025.
Part 500 requires New York-licensed insurers torun a risk-based cybersecurity program with a CISO, board oversight, encryptionof nonpublic information, third-party provider policies, 72-hour incidentnotice and annual certification. Since November 1, 2025, it also mandatesmulti-factor authentication for anyone accessing any information system and adocumented asset inventory.
According to the NAIC's state implementationreport as of March 2026, 28 jurisdictions have adopted the Insurance DataSecurity Model Law (#668). It requires a written information security program,a risk assessment, third-party oversight and notice to the insurancecommissioner after a cybersecurity event. Carriers in other states should stilltreat it as the baseline.
AI detects anomalies in logins, API calls anddata flows, spots AI-written phishing and voice deepfakes, and contains threatsautomatically by locking accounts or isolating systems. IBM's 2026 study foundthat organizations using security AI and automation cut breach costs by almost$2 million on average, which matters for carriers short on security staff.
Zero Trust, defined in NIST SP 800-207, meansno user, device or service is trusted because of its network location; everyrequest is authenticated and authorized. For insurers it limits damage fromstolen agent or employee credentials, and it aligns closely with the MFA,access and asset inventory requirements in NY DFS Part 500.
Use masked data in every test run, encrypt datain transit and in staging, log every migration operation, revoke temporaryaccess right after cutover and reconcile migrated records against the source.Running several trial migrations before the final cutover, as in Decerto'sGenerali Group Poland project, exposes security and data quality gaps early.
Talk to Decerto about insurance cybersecurity
If you are a mid-tier US P&C carrier reviewing cybersecurity in insurance software across your core platform in 2026, the expensive mistakes tend to sit in two places: security bolted onto the integration layer after go-live, and data exposed during migration.
Decerto's IT Audit and Architecture Review maps security across your insurance platform, identifies gaps between your core systems and your security tools, flags the Part 500 and Model Law items that touch the platform, and ends with a phased remediation plan. You keep the document, whether or not we work together afterwards.
The same team built Warta's eAgent platform for more than 40,000 agents and migrated an acquired insurer's data into Generali Group Poland's systems over a single weekend. Decerto holds a 4.9/5 rating on Clutch and has delivered insurance systems in 9 countries; see our mid-tier P&C carrier solutions. If a gap belongs to your endpoint, SIEM or network tools rather than the insurance platform, we will tell you that directly.
Sources
- IBM Security and Ponemon Institute. "Cost of a Data Breach Report 2026" and press release "IBM Study: One in Four Malicious Breaches are AI-Enabled, Costing Companies $6 Million on Average." July 29, 2026.
- Verizon Business. "2026 Data Breach Investigations Report - Executive Summary." May 2026.
- U.S. Department of Health and Human Services, Office for Civil Rights. "Change Healthcare Cybersecurity Incident Frequently Asked Questions." Last reviewed August 13, 2025.
- The Record by Recorded Future News. "UnitedHealth CEO confirms company paid $22 million ransom in heated Senate hearing." May 1, 2024.
- Office of the New York State Attorney General. "Attorney General James and DFS Superintendent Harris Secure $11.3 Million from Auto Insurance Companies over Data Breaches." November 25, 2024.
- Insurance Journal. "8 Auto Insurance Providers to Pay New York $19M Over Data Breaches." October 15, 2025.
- New York State Department of Financial Services. "23 NYCRR Part 500 - Cybersecurity Requirements for Financial Services Companies (Second Amendment)." November 2023, fully effective November 1, 2025.
- National Association of Insurance Commissioners. "State Implementation Report of NAIC-Adopted Model Laws and Regulations (as of March 9, 2026)." March 25, 2026.
- National Association of Insurance Commissioners. "NAIC Model Bulletin: Use of Artificial Intelligence Systems by Insurers." December 4, 2023.
- U.S. Department of Health and Human Services. "Summary of the HIPAA Security Rule."
- The Record by Recorded Future News. "Ransomware payments dropped in 2025 as attack numbers reached record levels: Chainalysis." February 26, 2026.
- Federal Bureau of Investigation, Internet Crime Complaint Center (IC3). "2025 IC3 Annual Report." April 7, 2026.
- National Institute of Standards and Technology. "SP 800-207: Zero Trust Architecture." August 2020.
- ISC2. "A Focus on Skills: The 2025 ISC2 Cybersecurity Workforce Study." December 4, 2025.
.avif)
.avif)




