Cyber policies must keep pace with a surge in incidents fueled by AI — the technology was a factor in a quarter of cyber breaches in the first half of 2026, according to Swiss Re, which estimates that 40% of large firms are underinsured.
Separately, business email compromise claims rose 57% in the second quarter of 2026 compared to a year earlier, with scammers focused more on account takeovers instead of expanding the role of ransomware, according to Travelers.
These threats are among those putting a strain on existing coverage limits, while opening up new areas for growth such as SMEs, most of which don't have cyber polices at all, according to Swiss Re.
Read more in Digital Insurance's recent coverage below:
40% of large firms underinsured as AI fuels cyber claims
One in four cyber breaches were AI-enabled in the first half of 2026, driving claims frequency and severity across the market, according to Swiss Re's latest cyber report. Large U.S. corporates average $120 million in coverage limits, but 10 losses per year over the past five years have exceeded that threshold. Forty percent of large firms were underinsured, but the SME segment represents the market's largest growth opportunity: Only 5%-20% of micro-SMEs and SMEs carry cyber policies, yet together they are projected to generate nearly $5 billion in premium by year-end. Global cyber market premium is forecast to reach $16.4 billion in 2026. Swiss Re warns that as AI amplifies ransomware, data breach and business interruption exposures, policy clarity on AI-related claims will be critical.
Read more:
BEC claims surge 57%; token theft drives account takeovers
Business email compromise claims jumped 57% in Q2 2026 year over year, driven largely by attackers stealing authentication tokens rather than passwords, according to Travelers' Q2 2026 Cyber Threat Report. Ransomware claims dipped 5% from Q1 as fraudsters shifted tactics. Phishing kits augmented by AI are increasingly used to hijack enterprise platforms such as Microsoft 365 and Google Workspace. Travelers recommends treating identity — not devices — as the primary unit of compromise: monitor for unexpected token issuance, new application consents and anomalous sign-ins, and build token-specific response protocols into incident plans rather than defaulting to password resets alone.
Read more:
How Banner Life hit No. 1 with AI and human oversight
Banner Life's rise to the largest U.S. term life insurer — 13% market share in Q1 2026 and $239 million in annualized premiums — offers a replicable blueprint: consolidate data first, then layer in AI. The carrier cut costs per application 27% over three years and dropped overall underwriting and operations expenses 23% by end of 2025, projecting another 16% reduction by year-end. Instant underwriting decisions grew from 67 to 109 per underwriter monthly. CEO Mark Holweger credits mandatory human oversight at every rule-implementation stage and early, continuous customer input — warning that excluding end users produces systems that "don't work how people expected."
Read more:
HappyRobot CEO: Stay customer-focused, not competitor-focused
HappyRobot, a voice-AI platform that automates high-volume operational workflows across calls, emails and documents, has raised approximately $200 million in total funding, including a $44 million Series B in 2025. CEO Pablo Palafox warns insurtech leaders against leading with technology rather than solving concrete workflow problems — a critical misstep in a regulated, high-stakes industry. For insurers, the near-term opportunity lies in deploying AI against manual coordination bottlenecks — intake, follow-ups, document collection — and demonstrating measurable ROI in production environments. Palafox predicts that AI will shift from task assistance to managing entire workflows within five years, with human oversight reserved for exceptions and complex judgment calls.
Read more:
Outdated core systems are blocking AI scale for insurers
The gap between AI ambition and execution often traces back to legacy Policy Administration Systems, inconsistent data and layered AI tools that fall out of sync. Applying Amdahl's law, a system's speed is constrained by its slowest component — for most insurers, that's the PAS. Stacking AI agents on a flawed foundation doesn't fix inconsistencies; it accelerates them. Smaller carriers and MGAs face additional resistance due to budget constraints, but practical, workflow-integrated use cases can build momentum. Prioritize clean data, unified product logic and ISO 27001:2022-certified platforms before expanding AI deployment. Human review of AI outputs remains essential to catch hallucinations and data integrity issues.
Read more:
This roundup was created with AI assistance. A Digital Insurance editor reviewed each item before publication.
