Admiral fraud claims jump 71% as AI image manipulation spreads

TL;DR

• Cardiff-based Admiral reported a 71% rise in insurance fraud during 2025, partly attributing the jump to AI image manipulation and document fabrication
• Detected AI cases included fake gold-and-diamond watches, exaggerated car damage, and repositioned number plates used for duplicate claims
• The Insurance Fraud Bureau says the industry is “heavily concerned” and investing in AI-detection countermeasures — an arms race that UK consumers will ultimately fund through premiums

Admiral, the Cardiff-headquartered motor and home insurer, has reported a 71% rise in fraud during 2025 compared to the previous year, with generative AI tools featuring in a growing proportion of cases. BBC Wales reported on the detection work, including a clearly AI-generated image of a gold-and-diamond watch submitted as part of a stolen-items claim, manipulated photos showing exaggerated car damage, and a case where a number plate was digitally altered and repositioned to create a duplicate claim.

The new fraud pattern

Fraudulent AI-generated evidence breaks down into two categories by perpetrator. Admiral’s fraud team describes “opportunistic” customers using widely available image-editing tools to exaggerate otherwise genuine claims — adding damage to real photos, inventing missing items. The second, more serious category involves organised crime gangs using AI to fabricate entire documents at scale, making large coordinated fraud attempts more efficient than manual forgery.

John Davies of the Insurance Fraud Bureau told the BBC the industry is “heavily concerned” and “investing in technology.” The detection tooling is also AI-based: image authenticity classifiers, metadata analysis, and cross-claim pattern detection. All the cases highlighted in the BBC report were caught, with claims rejected and policyholders facing potential prosecution.

Why this matters for UK consumers

The immediate cost falls on honest policyholders. UK motor and home insurance premiums have already risen sharply over the past three years; fraud costs are consistently cited in industry briefings as a significant contributor to premium inflation. A 71% single-year rise at one of the UK’s largest insurers suggests AI has meaningfully changed the fraud economics — lowering the cost and technical barrier to attempt at volume.

The counter-investment pattern mirrors banking: insurers deploying AI to detect AI-generated attacks. This is the same arms-race dynamic NCSC CEO Richard Horne highlighted this week for cyber defence — AI accelerates both sides of the threat landscape simultaneously, and weak defenders get exposed fastest. For insurers, the defender question is whether fraud detection investment keeps pace with offensive tool availability; so far, Admiral’s detection wins suggest yes, but the trajectory is the real story.

Looking forward

Expect the Association of British Insurers to push harder on consumer education campaigns, harsher cancellation and prosecution outcomes for AI-assisted fraud, and potentially regulatory movement on provenance requirements for claim evidence. The ICO and Ofcom may have roles to play if AI-image tools begin marketing themselves with capabilities tailored to evidence fabrication. For UK policyholders: premium impact is already landing. The honest question is how much of the 2025 increase trace back to AI-assisted fraud, and whether that pressure eases or accelerates in 2026.