Introduction
Fraud in property and casualty insurance is no longer a background issue, it’s a front-line business risk.
Recent figures show the scale of the problem:
- UK insurers detected £1.1bn in fraudulent claims in 2023 across around 84,400 cases. [1]
- In the US, insurance fraud is estimated to cost $308.6bn annually, adding roughly $900 to each policyholder’s bill. [2]
- In Australia, insurance fraud is estimated at up to AU$2.2bn a year, with insurers also facing billions in reported scam losses. [3]
In Australia alone, ICA members identified around AU$560m in opportunistic fraud in 2023, with a further AU$400m thought to go undetected every year. [3]
So whether you’re in the UK, Australia or the USA, the story is similar: higher costs, more online business and more scams are putting extra pressure on fraud teams.
This guide summarises the key fraud challenges and trends heading into 2026 and sets out practical priorities for fraud leaders.
1. The fraud landscape heading into 2026
Data from the last couple of years points clearly to where fraud is heading next year:
- More pressure on honest customers and more temptation to commit fraud. Cost-of-living pressures and rising premiums are driving opportunistic behaviour, from exaggerated losses to non-disclosure at application. (1)
- Scams and financial crime bleeding into P&C. Large-scale scam losses and new regulation (like Australia’s Scams Prevention Framework) mean insurers are expected to play a bigger role in detecting and disrupting scams that involve policies and claims. (5)
- Organised and cross-border activity. Staged accidents, identity theft and policy farming are being seen across multiple markets, often involving the same patterns and enablers. (6)
These shifts set the context for the specific challenges fraud teams face in 2026.
2. Key challenges for P&C fraud teams
2.1 Rising volume and case complexity
Fraud teams will continue to see:
- More claims with fraud indicators, from opportunistic inflation to fully staged events.
- More information per case (telematics, images, OSINT, devices) to review under tight SLAs.
- Evidence that fraud is contributing to higher motor and property claims costs, such as Australian motor premiums rising significantly since 2019. (8)
The result is workloads rise faster than headcount, increasing the risk of both missed fraud and poor customer experience.
2.2 Application, identity and synthetic identity fraud
Front-end fraud is now as critical as claims fraud:
- UK insurers are stopping large volumes of fraudulent applications at point of sale, typically driven by misrepresented driving history, parking and mileage. (1)
- In one Australian motor survey, nearly 7% of policyholders admitted providing incorrect information to reduce their premium. (8)
- Synthetic identity fraud, blending real and fake personal data to create a ‘new’ identity, is now recognised as a major driver of new-account fraud globally. (14)
For insurers, this shows up as:
2.3 Staged accidents and policy farming
Traditional P&C fraud typologies are evolving:
- US and international guidance still highlights staged accident patterns (swoop and squat, drive-down, side-swipe) as a major exposure for motor insurers.
- In the UK, crash-for-cash gangs now use mopeds to engineer low-speed collisions and harvest licence and policy data for later misuse. (7)
- In Australia, a New South Wales case involving 16 collisions and 45 vehicles showed how staged accidents, false damage, third-party cover and identity abuse can be combined into a profitable scheme. (8)
These schemes are often part of wider policy farming, opening multiple policies using real or synthetic identities to claim against later. That demands network and entity analytics, not just case-by-case rules.
2.4 Deepfakes, “shallowfakes” and AI-enabled evidence
AI should be recognised as firmly part of the fraud toolkit:
- Swiss Re’s SONAR report flags deepfakes and disinformation as emerging risks, including their use to support fraudulent claims and cyber attacks. (9)
- Insurers are seeing rapid growth in ‘shallowfake’ images – photos and documents manipulated with simple apps to exaggerate or invent damage.(9)
Fraud and claims teams increasingly need:
- Automated image and document forensics (metadata checks, noise and pattern analysis, layout anomalies);
- Clear policies on when to challenge questionable media.
- Training so investigators can recognise common signs of AI manipulation.
2.5 Regulation, governance and operational risk
AI and fraud controls sit squarely in regulators’ sights:
- The EU AI Act and EIOPA guidance treat many insurance AI use cases, including claims triage and fraud scoring, as high-risk, with expectations on documentation, monitoring and human oversight. (11)
- EIOPA also stresses the need for explainability, fairness and data quality, which applies directly to fraud models and automated decisions. (11)
- In Australia, APRA’s CPS 230 reframes fraud and scams as part of operational risk and resilience, requiring boards to understand and control AI-driven systems and key third-party providers. (12)
- The Scams Prevention Framework adds further accountability for preventing and responding to scams that affect customers. (5)
For fraud leaders, this means detection isn’t just about catching more fraud; it’s about building auditable, well-governed systems that stand up to regulatory and legal scrutiny.
2.6 Data silos and limited collaboration
Industry surveys consistently highlight the same pain points:
- Data is often fragmented across quote, policy, billing, claims and digital channels, limiting the ability to see the full customer and entity view.
- More than 65% of surveyed fraud professionals rank access to broader/shared data as a top priority for improving fraud strategy. (10)
- In Australia, the ICA is building dedicated counter-fraud and scams capabilities to share intelligence across members, reflecting the fact that fraud patterns regularly span multiple carriers. (13)
The challenge for individual insurers is to connect internal silos and plug into industry-level intelligence while staying compliant with privacy and data-sharing rules.
3. Strategic priorities for 2026
3.1 Build an ‘intelligence-first’ fraud data layer
Move from rule-based controls in individual systems to a unified intelligence layer that:
- Resolves entities (people, vehicles, addresses, devices, businesses) across policies and claims.
- Understands behavioural signals from digital journeys, payments and contact channels.
- Continuously feeds back confirmed fraud and false positives to improve models.
This is especially valuable for motor and property portfolios in markets like Australia, where large weather events create sudden surges in claims.
3.2 Strengthen identity controls and collaboration
Given the growth in identity and synthetic identity fraud:
- Apply multi-layered identity proofing at onboarding (documents, data sources, behavioural signals), with risk-based step-up.
- Combine identity, device and behavioural data to detect synthetic patterns, not just rely on KYC checks.
- Ensure fraud, underwriting and financial crime teams share tools, typologies and outcomes, so identity risk is managed consistently across the lifecycle.
- This is especially valuable for motor and property portfolios in markets like Australia, where large weather events create sudden surges in claims.
3.3 Use AI, but with governance and explainability
AI and machine learning are essential for modern fraud detection, but they must be under control:
- Maintain a clear model inventory for all fraud-relevant models, including image/document models, not just scores.
- Run regular back-testing, monitoring and fairness checks, and record decisions on model changes.
- Give investigators access to human-readable reasons behind scores so they can make defensible decisions and explain outcomes to customers and regulators.
- This is especially valuable for motor and property portfolios in markets like Australia, where large weather events create sudden surges in claims.
3.4 Invest in people and skills
Finally, tools only deliver value if people can use them:
- Build hybrid roles that combine investigative experience with data literacy, so teams can interpret model output and challenge it when needed.
- Create structured training on fraud typologies, OSINT, AI-enabled fraud and regulatory expectations.
4. What this means for fraud teams working with kbs Intelligence
Our fraud detection software can help you to:
- Connect data across policies, claims and external sources so you can surface organised and identity-driven fraud that individual systems miss.
- Use advanced analytics to spot patterns and networks while keeping decisions transparent and explainable for investigators, risk teams and regulators.
- Put the right workflows and controls in place, with audit trails, role-based access and governance features that support CPS 230, AI rules and emerging scam-prevention expectations.
References
- [1] Association of British Insurers (ABI), industry fraud statistics for 2023 (e.g. detected £1.1bn fraudulent claims; c.84,400 cases).
- [2] Insurance Information Institute / Coalition Against Insurance Fraud, estimate that insurance fraud costs the US $308.6bn annually, adding around $900 per policyholder.
- [3] Insurance Council of Australia (ICA) and industry reporting on AU$560m detected opportunistic fraud and AU$400m estimated undetected fraud in 2023, mainly across motor and property.
- [4] Industry summaries of Australian insurance fraud, citing estimates of up to AU$2.2bn annually and ACCC data on scam losses.
- [5] ACCC and Australian Treasury material on scam losses and the development of the Scams Prevention
