Lauren France, member of the Forum of Insurance Lawyers (FOIL) and counter fraud director at DWF examines the rising risks as artificial intelligence (AI) becomes increasingly part of a rising tide of fraudulent claims.

Artificial Intelligence is something that has been around for many years. Over the past two years, however, there has been significant discussion around its use within the insurance space, as well as its deployment within the claim process in the furtherance of efficiency and improved capability. The opportunities are seemingly plentiful.

Yet there is a dark side to the seemingly plentiful opportunities when Generative AI is deployed in the wrong hands. The deliberate misuse of AI is widespread and it is here that we have seen AI rapidly become an emerging fraud risk in the insurance world.

Recent research by Cifas found that last year a record number of scams fuelled by AI – 444,000 – were reported to the national fraud database in 2025, a 6% increase on 2024.

There has been a real explosion of Generative AI in the past 18 months. Its capabilities have developed enormously, and it is now easily accessible. In the past, fraudsters have leveraged technology, such as photo editing tools, to support fraudulent claims by manually and deliberately manipulating images. These are known as ‘Shallow Fakes’.

The advancements in Generative AI creates even further opportunity for those who want to misuse it for financial gain. This type of AI does what it says on the tin; it creates new content based on direct prompts. It is capable of creating entirely new images, video, audio, and text. It can also be used to alter existing images; for example, by adding damage to a particular item or image on demand.

Equally, there has been a substantial rise in instances of false identification documents being submitted in claims.

Some examples include the exaggeration of bruising in a personal injury claim, additional physical damage to a vehicle in a Road Traffic Accident claim, fake water damage overlaid onto a perfectly clean image, fake driver’s licences, and fake damages to items such as iPads and laptops.

Large Language Models (LLMs), a type of Generative AI, generate human-like language and are capable of creating entire documents in a matter of seconds. This type of AI is responsible for the ‘hallucinations’ that we have seen reported in recent months. It is easy to see how this could enhance the approach of litigants in person in claims.  LLMs could create and lend more credence to documents such as pleadings and witness statements, giving further weight to an individual claim. The availability of these tools may therefore encourage litigants in person to take greater risks in presenting claims. The result is a likely increase in the volume of borderline claims that are not necessarily sophisticated and which lack merit, but which on the surface are entirely plausible and not easily detectable.

Another recent example was seen in the case of UAB Business Enterprise v Oneta Ltd & Ors [2026] EWCH 543 (Ch), in which a witness made use of  “smart glasses” whilst giving evidence. Thankfully, that deception was detected and the witness found to have been both untruthful and clearly deliberately coached.

AI-enabled fabrication risks add to costs in two significant ways: more claims requiring enhanced verification and validation, and greater leakage where fabricated evidence succeeds and goes undetected. Both in turn can drive an increase in premiums.  Even where a claim is successfully repudiated, an insurer may have incurred increased handling costs, investigative spend, and even litigation costs.  It is important to recognise that AI-enabled fraud is not only confined to individual claims and outcomes, but naturally will also affect wider economics. Relevant to this point is investment in technology by insurers – using AI to beat AI. This too comes at an increased cost.

We are in an era where it is possible to create realistic synthetic images and videos in a very short amount of time, and with minimal effort. It is important that as an industry we are ready to respond.   A key point for legal professionals and insurers is to not accept any document, image, or video at face value and without increased scrutiny.  In addition to the use of technology being used to validate any suspicious documents, manual validation and traditional investigations remain key and should be at the forefront of any claims handling. For example, physical inspection of any particular item where manipulated evidence is suspected is crucial, along with utilising a reliable secondary source to validate any documents where that option is available.