← Blog
explainers·Nov 19, 2026·10 min read

Insurance Fraud Statistics 2026: AI, Deepfakes and What Fraud Costs

Insurance fraud statistics for 2026 from named primary sources: US and UK fraud costs, how AI and deepfakes are changing claims fraud, detection accuracy and the outlook.

ShareLink copied
Insurance Fraud Statistics 2026: AI, Deepfakes and What Fraud Costs

Most "AI insurance fraud" statistics online have no source. Every number on this page links to a named primary source. Where insurance-specific data doesn't exist yet, we say so and use the closest reliable proxy.

Last updated November 19, 2026.

AI insurance fraud in four numbers
AI insurance fraud in four numbers

Key Insurance Fraud Statistics

StatisticFigureSource
US insurance fraud, all lines~$308.6B per yearCoalition Against Insurance Fraud[1]
Share of US P&C losses involving fraud~10%Coalition Against Insurance Fraud[2]
UK fraudulent claims detected, 2025£1.34bn, up 14%Association of British Insurers[3]
Zurich UK fraudulent claims detected, last 12 months4,359 claims, £109.3m (one every two hours)Zurich UK[4]
Insurers saying AI editing tools drive more fraud98%Verisk, 2026[5]
Insurers very confident they can spot a deepfake32%Verisk, 2026[5]
Rise in synthetic voice attacks on insurers, 2024475%Pindrop, 2025[6]
Anti-fraud teams more than moderately prepared for AI fraud7%ACFE and SAS, 2026[7]
Human accuracy identifying AI-generated faces48.2% (chance is 50%)PNAS, 2022[8]
Gen-AI-enabled US fraud losses forecast for 2027$40BDeloitte, 2024[9]

How Big Is Insurance Fraud?

Where US insurance fraud losses sit
Where US insurance fraud losses sit

United States

The Coalition Against Insurance Fraud estimates US insurance fraud at $308.6 billion a year, including about $45 billion in property and casualty, $34 billion in workers' compensation and $7.4 billion in auto theft.[1] Fraud is present in roughly 10% of P&C losses.[2]

United Kingdom

Insurers detected £1.34 billion of fraudulent claims in 2025, up 14% on the year even as case numbers fell, including £625 million across 51,900 motor claims.[3] Zurich UK alone detected 4,359 fraudulent claims worth £109.3 million in 12 months, its highest ever, and named generative AI among the tools fraud gangs now use.[4]

Australia

Insurance Council members detected about A$560 million of opportunistic motor and property fraud in 2023, and estimate about A$400 million more goes undetected each year.[10]

Insurance fraud was a large problem before AI arrived
Insurance fraud was a large problem before AI arrived

None of these figures separate out AI-generated evidence. That data doesn't exist yet at industry level.

How AI Is Changing Insurance Fraud

What insurers report

  • Verisk (2026): 98% of 300 US claims professionals agree AI editing tools are driving more digital insurance fraud. Only 32% are very confident they could identify a deepfake.[5]
  • Zurich UK (2026): lists AI-generated documents and evidence alongside staged accidents and ghost broking as tactics used by organized fraud networks.[4]
  • Swiss Re Institute (2025): rates deepfakes and disinformation as a high-impact emerging risk for insurance fraud.[11]
  • ACFE and SAS (2026): 44% of anti-fraud professionals saw deepfake social engineering rise significantly in the past two years. Only 7% feel more than moderately prepared.[7]

Signals from adjacent sectors

The clearest signals come from adjacent sectors
The clearest signals come from adjacent sectors
  • Pindrop: deepfake fraud attempts rose more than 1,300% in 2024. Synthetic voice attacks on insurers rose 475%, the fastest of any sector it tracks.[6]
  • Sumsub: deepfakes detected in identity checks rose 4x from 2023 to 2024.[12]
  • Signicat: deepfakes now make up about 6.5% of fraud attempts at European financial institutions, including insurers.[13]
  • FBI IC3: reported internet crime losses reached $20.9 billion in 2025, up 26%.[14]

Can People and Machines Spot AI Fakes?

  • People can't. In a peer-reviewed PNAS study, 315 participants identified AI-generated faces with 48.2% accuracy, slightly worse than a coin flip.[8]
  • Machines do better, but less than lab numbers suggest. The Deepfake-Eval-2024 benchmark found detector performance fell by 45% to 50% on real-world deepfakes compared with older academic benchmarks.[15]

What AI-Enabled Fraud Could Cost

Generative AI could more than triple US fraud losses by 2027
Generative AI could more than triple US fraud losses by 2027

Deloitte projects generative-AI-enabled fraud losses in US financial services could reach $40 billion by 2027, up from $12.3 billion in 2023, a 32% annual growth rate.[9] That forecast covers financial services broadly, not insurance alone.

No one has measured what AI-made evidence adds to insurance fraud yet. A simple sensitivity model shows the scale:

What AI-made evidence could add to US P&C fraud
What AI-made evidence could add to US P&C fraud

Illustrative model, not a measurement. If AI-made or AI-altered evidence adds 5% to 20% to the ~$45 billion US P&C fraud baseline, that is roughly $2.3 to $9.0 billion a year. The middle case is plausible because most fraud tools score claim data rather than media authenticity, digital-first claims mean fewer inspections, and fakes keep getting cheaper.

The payout is only part of the cost. Investigation, legal work, reviewer hours, delays to honest claims and reputational damage all add to it, and honest policyholders pay for all of it through premiums.

Regulation at a Glance

JurisdictionRuleStatus
USNAIC Model Bulletin on insurers' use of AIIn force in 24 states plus DC (April 2026)[16]
USFinCEN alert FIN-2024-Alert004 on deepfake fraudIssued November 13, 2024[17]
EUAI Act Article 50 deepfake transparency rulesApplies from August 2, 2026[18]

No regulator has yet issued insurance-specific guidance on AI-generated claim evidence.

Outlook

How deepfake fraud in insurance is likely to develop
How deepfake fraud in insurance is likely to develop

How Witness Helps

Witness checks photos and video for signs of AI generation and editing. Each scan returns a clear verdict, Real or AI, with a confidence score and a plain-language explanation.

Try a scan or see Witness for insurance.

Frequently Asked Questions

How much does insurance fraud cost each year?

About $308.6 billion a year in the US across all lines, according to the Coalition Against Insurance Fraud. UK insurers detected £1.34 billion of fraudulent claims in 2025.

What percentage of insurance claims are fraudulent?

The Coalition Against Insurance Fraud estimates fraud is present in about 10% of US property and casualty losses.

How much insurance fraud involves AI or deepfakes?

No regulator or industry body reports it separately yet. Nearly all insurers believe AI editing is driving more fraud, and adjacent sectors show deepfake attacks growing several-fold per year.

Can people spot AI-generated images?

Not reliably. In a PNAS study, people identified AI-generated faces at about chance level.

◆◆◆

Related reading

Verisk, Pindrop, Sumsub and Signicat sell fraud-prevention products. Their figures are self-reported.

  1. [1]Coalition Against Insurance Fraud, The Impact of Insurance Fraud on the U.S. Economy, August 2022.
  2. [2]Coalition Against Insurance Fraud, Fraud Stats.
  3. [3]Association of British Insurers, Fraudsters target bigger payouts as insurers detect £1.34 billion of bogus claims, October 2026.
  4. [4]Zurich UK, Zurich detects £109m of fraud, November 5, 2026.
  5. [5]Verisk, AI Editing Tools Are Fueling a New Era of Insurance Fraud, March 17, 2026.
  6. [6]Pindrop, 2025 Voice Intelligence & Security Report, June 12, 2025.
  7. [7]ACFE and SAS, 2026 Anti-Fraud Technology Benchmarking Report, March 2026.
  8. [8]Nightingale and Farid, AI-synthesized faces are indistinguishable from real faces and more trustworthy, PNAS, 2022.
  9. [9]Deloitte Center for Financial Services, Generative AI is expected to magnify the risk of deepfakes and other fraud in banking, May 29, 2024.
  10. [10]Insurance Council of Australia, Insurance industry boosts fight against fraud and scams, May 21, 2024.
  11. [11]Swiss Re Institute, SONAR 2025: How deepfakes, disinformation and AI amplify insurance fraud, June 2025.
  12. [12]Sumsub, Identity Fraud Report 2024, November 2024.
  13. [13]Signicat, The Battle Against AI-driven Identity Fraud, February 2025.
  14. [14]FBI Internet Crime Complaint Center, 2025 Internet Crime Report, April 2026.
  15. [15]Chandra et al., Deepfake-Eval-2024, arXiv, 2025.
  16. [16]NAIC, Implementation map of the Model Bulletin on the Use of AI Systems by Insurers, April 2026.
  17. [17]FinCEN, FinCEN Issues Alert on Fraud Schemes Involving Deepfake Media, November 13, 2024.
  18. [18]EU AI Act, Regulation (EU) 2024/1689; European Parliamentary Research Service, AI Omnibus briefing782651_EN.pdf), 2026.
Share this articleLink copied

Scan images while you browse.

The Witness Chrome extension checks any image on any website. Right-click to scan.

Add to Chrome

Check a file yourself.

Try 5 scans free now. Sign up for 20 a month.