In 2024, a tradesman posted a photo of his van online to promote his business. Fraudsters copied the image, edited a crack into the front bumper, and submitted a motor claim in his name for an accident that never happened. They even included a repair invoice for more than £1,000.

At first glance, the claim looked genuine. It was only when LV's fraud team compared the submitted photo with the original image on social media that they uncovered the deception. The two images were identical apart from the added damage. The case was reported by The Guardian in May 2024 and later confirmed by Allianz.

What makes this case significant is how ordinary it appeared. The invoice looked legitimate. Even the photo looked real until it had been altered. Nothing about the submission would have raised concerns during a routine document check. That is the challenge insurers are now dealing with.

The problem is also growing. The ABI reported £1.16 billion in detected fraudulent general insurance claims in 2024, with more than 98,400 cases uncovered, a 12% increase on the previous year. And those figures only reflect the fraud that was detected.

Fraud has become easier to scale

Generative AI has not created insurance fraud. Inflated or exaggerated claims remain the largest category, accounting for £466 million in detected fraud during 2024 according to the ABI.

What has changed is how easy it has become to produce convincing evidence. Editing damage into a photo once required specialist software and experience. Today it can be done in minutes with a mobile app or just using a simple prompt.

Allianz and Zurich reported a 300% increase in manipulated images, often referred to as shallowfakes, between 2021 to 2022 and 2022 to 2023. Shift Technology also estimates that 20 to 30% of claims may now contain altered images, fabricated documents or synthetic reports.

The same pattern appears across different types of evidence. Photos of vehicle damage can be edited. Salvage listings can have registration plates changed. Invoices can be generated instead of issued. Synthetic identities can be built using breached personal data.

Each piece of evidence may appear credible on its own. The problem only becomes visible when it is viewed alongside other claims and historical records.

Detection is only part of the answer

Insurers are investing in image forensics, metadata analysis and voice analytics, and these tools are producing results. Allianz reported detecting 29% more fraud in 2023 after strengthening its controls.

The challenge is that detection will always be reactive. As technology improves the ability to identify fake evidence, it also improves the ability to create it, often at a lower cost.

Fraud rarely reveals itself through a single document. A manipulated photo may sit in one claim. A reused invoice may appear in another. A suspicious pattern may only emerge across several policies handled by different teams.

When claims data sits in disconnected systems, those connections are much harder to spot.

Why connected claims data matters

This shifts the conversation away from individual fraud tools and towards the claims platform itself. Every claim, payment and supporting document should form part of a connected record.

The capabilities that matter are already available in modern claims platforms. Evidence can be captured directly through a claims app with its metadata preserved instead of being uploaded later from an unknown source. Claims data can be structured so photos, invoices and other supporting documents are automatically checked against previous claims. Audit trails can show who accessed or changed information and when. Integration with organisations such as the Insurance Fraud Bureau also supports wider data sharing across the industry.

These are not future capabilities. They are practical ways insurers can improve fraud detection today.

A fake claim is much easier to hide when information is fragmented. It becomes much easier to identify when every part of the claim can be viewed in context.

Success is not just about catching fraud

Most claims are genuine. A fraud strategy that treats every customer with suspicion creates unnecessary delays for honest policyholders. Under Consumer Duty, that is not only poor customer service but also a regulatory risk.

The real measure of success is not simply how much fraud an insurer detects. It is how quickly genuine claims can be verified and settled.

That is why Allianz has said its investment in fraud controls is intended to achieve two outcomes: detect more fraud and speed up legitimate claims. Those objectives support each other. When evidence can be verified as it enters the claims process, fewer genuine claims need to be held up while suspicious cases are investigated.

Another shift is also taking place. Instead of asking whether evidence looks fake, insurers increasingly need to ask whether its origin can be verified. As AI generated content becomes more convincing, knowing where evidence came from becomes more valuable than trying to judge how authentic it appears.

Fraud will continue to evolve, and so will the technology used to detect it. The insurers best prepared for that future will be those with connected claims data that is complete, verifiable and easy to trust.