AI Arms Race, Systemic Risk, and a Fraud Bill That's Still Climbing

Cyber, AI, and Economic Pressure Are Converging: RiskScan 2026 Maps the New Risk Landscape

The Insurance Information Institute and Munich Re US published RiskScan 2026 this week, a cross-market study drawing on more than 1,700 respondents across five insurance market segments in the US and UK. The headline finding is that today’s risk landscape is no longer defined by isolated threats. Cyber incidents, economic volatility, AI, natural catastrophes, business interruption, and emerging liability exposures are increasingly arriving in clusters, reinforcing one another in ways that complicate both underwriting and risk management planning.

AI ranked as the most impactful emerging technology in the survey, reflecting both the pace of adoption and the breadth of concerns it generates, including operational, regulatory, liability, and systemic dimensions. That range of concerns matters for how risk managers and carriers think about AI exposure. It is not simply a technology risk or a cyber-adjacent risk. It surfaces across governance, operations, and accumulation scenarios that existing frameworks were not built to address simultaneously.

The study also highlights persistent protection gaps in flood and cyber insurance, a finding that has direct relevance for brokers navigating client conversations about coverage adequacy. Growing recognition of legal system abuse as a structural driver of P&C costs rounds out the picture, pointing to the social inflation dynamic that has been compressing combined ratios across casualty lines.

The specialty insurance component of the report focuses on the tight interconnection between cyber incidents, business interruption, new technologies, and natural catastrophes. The implication for accumulation modeling is significant. Events that would historically have been treated as independent perils are now more likely to trigger correlated losses, particularly where AI infrastructure or cloud dependencies are involved. For reinsurers and carriers pricing catastrophe programs, that interdependency is not yet well-reflected in most models.

Marcus Winter, president and CEO of North America P&C Re at Munich Re US, noted the industry’s commitment to helping clients understand and manage increasingly complex risk, while Kerri Hamm, Munich Re US’s head of cyber underwriting, pointed to the need to accelerate product development in the areas the survey identifies as most pressing for P&C clients and policyholders.

— Kane Wells at Reinsurance News

URL: https://www.reinsurancene.ws/cyber-ai-and-economic-pressures-dominate-insurance-concerns-in-riskscan-2026/

SAS: The Barrier to Fraud Is Gone. Anyone With a Computer Can Fake a Claim.

The Aviva numbers reflect a UK market reality, but SAS published research this week documenting the same dynamic at a structural level applicable across any market where digital image submission is standard practice in claims handling. The central argument is straightforward: generative AI tools have eliminated the technical skill requirement for fabricating convincing claims evidence. What previously required access to photo-editing expertise or professional fraud networks can now be accomplished in seconds by anyone with an internet connection.

Insurance fraud in the United States costs consumers an estimated $308.6 billion annually, with roughly one in ten P&C losses already involving some element of fraud. SAS conducted a live demonstration through fraud specialist Adam Hall showing how AI tools could produce believable crash scenes almost instantly, closely mirroring documented tactics already in use by organized crime groups targeting US and European carriers. The demonstration was designed to illustrate not a theoretical future scenario but the current state of capability available off-the-shelf.

The report’s implications go beyond the immediate fraud threat. Verisk’s 2026 State of Insurance Fraud Report found that 36 percent of surveyed consumers indicated a willingness to submit digitally manipulated photos in support of an insurance claim, a figure that represents a meaningful shift in fraud tolerance compared to prior research. When the barrier to fraud drops and social tolerance simultaneously rises, the claims environment changes in ways that pricing models built on historical loss data are not designed to capture.

Franklin Manchester, principal global insurance advisor at SAS, was direct about the dual role AI is playing: the same technology enabling fraud fabrication can detect anomalies in images that human adjusters cannot see. Synthetic image detection tools are now deployable at the claim intake stage, providing a countermeasure that operates at the same speed and scale as the threat. Carriers that have not yet invested in AI-assisted fraud detection at the image level are working with a structural disadvantage that is measurable in claims leakage.

— Jonalyn Cueto at Insurance Business

URL: https://www.insurancebusinessmag.com/us/news/cyber/warning-over-new-ai-insurance-scams-577092.aspx

Aviva’s Fraud Bill Hit a Record Last Year. Generative AI Is Why.

Aviva disclosed this week that it detected a record £233 million in fraudulent and suspect claims across its portfolio in 2025, the first full year incorporating the Direct Line brands it acquired the prior summer. Over 18,400 suspect claims were identified. The nature of the fraud, not just the volume, is what makes this disclosure significant.

Fraudsters are no longer primarily relying on staged collisions or inflated repair invoices assembled with the help of rogue garages or complicit medical professionals. Generative AI now allows individuals or small groups to manufacture convincing supporting evidence, fake accident scenes, fabricated repair documents, exaggerated damage images, without ever leaving their desk or requiring access to corrupt professional networks. Motor insurance accounted for more than seven in ten detected fraud cases. The value of motor fraud was up 39 percent year over year, with fraudsters shifting from collision staging toward higher-value exaggerated damage and injury claims backed by AI-generated documentation.

The pattern extended into liability and travel lines. While claim counts in liability were broadly stable, the average value of fraudulent claims rose 32 percent, with claimants using AI tools to exaggerate loss-of-earnings calculations, rehabilitation costs, and injury severity. Pete Ward, Aviva’s head of claims counter-fraud, noted that professional enablers, including rogue lawyers and medical professionals, are amplifying the problem by lending legitimacy to claims built on fabricated evidence.

Aviva’s response is to fight AI fraud with AI detection. The insurer has deployed its own analytical systems to flag suspicious patterns at scale, cross-referencing image metadata, claim physics, repair cost benchmarks, and network relationships between claimants. The architecture is kept under human oversight, but the scale of processing required for pattern detection means automation is now a prerequisite rather than a supplement. The case Aviva cited as emblematic involved a staged collision where fraudsters sought £470,000 in claims, ultimately resulting in criminal convictions after court video confirmed no witnesses had been present at the scene.

— Business Insurance Team at Business Insurance

URL: https://www.businessinsurance.com/aviva-fights-ai-fraud-with-ai-of-its-own/

This Week’s TL;DR

The theme running through this week’s coverage is not AI as a future risk to be planned for. It is AI as a present-tense operational condition that is reshaping fraud economics, accumulation modeling, and market-wide risk perception simultaneously. RiskScan 2026 provides the macro frame: cyber, economic pressure, AI, and natural catastrophe are no longer arriving as separate perils with independent frequency and severity. They are interconnected exposures that reinforce each other, creating loss scenarios that current aggregation models were not designed to price. The protection gaps the study highlights in flood and cyber are not simply market failures; they are indicators of a pricing and modeling infrastructure that has not yet caught up with how risks now correlate.

Aviva’s fraud disclosure puts concrete numbers on the operational side of that problem. A record £233 million in detected fraud, driven largely by generative AI tools enabling evidence fabrication at scale, is not a data point about one carrier’s claims department. It is a signal about what happens to fraud economics when the barrier to producing convincing false evidence collapses. The shift from staged collisions toward AI-documented exaggerated claims, and the 39 percent rise in motor fraud value, reflects fraudsters rationally adapting to the tools available. The professional enabler dimension adds further complexity: when rogue lawyers and medical professionals combine with AI-generated documentation, the claims that result are harder to detect and more expensive to litigate.

SAS sharpens the implications by documenting that this is not a specialized capability. The same tools producing Aviva’s fraud problem are available to any consumer with a laptop, and a third of US consumers surveyed indicated willingness to use them. For carriers, that statistic has underwriting and loss-development implications that have not yet worked through to reserve adequacy or pricing models. The countermeasure that SAS documents, AI-assisted image analysis at the point of claim intake, is deployable now, but uptake across the market is uneven, and carriers that have not yet invested are absorbing losses that better-equipped peers are intercepting.

For risk managers, the week’s coverage describes an environment where the risk landscape is more interconnected, fraud is more technically accessible, and the detection gap between AI-enabled attack and AI-enabled defense is the variable that will separate better-performing carriers from laggards over the next several renewal cycles.