— Insurance companies face growing challenges distinguishing legitimate claims from those supported by AI-generated photographs and manipulated documents. As synthetic images become more convincing, fraud investigators are being encouraged to examine multiple sources of evidence rather than rely exclusively on automated image detection.

The threat is already affecting claims operations. British insurer Aviva identified more than 18,400 suspected fraudulent claims valued at £233 million in 2025 across its brands, including Direct Line. The company reported an increase in AI-generated accident photographs and altered documents used to support false or exaggerated claims, particularly in motor insurance.

Marcin Nowak, a board member at insurance technology company Decerto, argues that detecting fabricated images requires a broader investigative approach. Writing for Forbes Technology Council, Nowak explained that image recognition systems can identify suspicious alterations but may struggle to keep pace with increasingly sophisticated generative AI models.

Instead, insurers can compare image metadata, policy histories, repair invoices and previously submitted claims to identify inconsistencies. Discrepancies involving reported loss dates, recent coverage changes or repeated damage photographs may justify additional investigation, although individual anomalies are insufficient to establish fraud.

Automated fraud detection also creates risks for legitimate policyholders. Systems that flag suspicious activity without learning from subsequent investigations can repeatedly identify innocent claims as potentially fraudulent. The result can be unnecessary payment delays, increased investigative expenses and additional work for claims adjusters.

Nowak recommends incorporating verified investigation outcomes into fraud detection models. When an SIU investigator determines that a flagged claim is legitimate, that result should inform future detection thresholds and help reduce recurring false positives.

Regulators are also examining insurers' use of AI in consumer-facing decisions. The National Association of Insurance Commissioners' 2023 model bulletin specifically addresses fraud detection and establishes expectations for governance, monitoring and documentation of AI systems used by insurers. The guidance emphasizes compliance with existing insurance laws, including protections against unfair discrimination.

For claims departments, effective fraud prevention increasingly depends on combining automated screening with experienced investigators who can evaluate conflicting evidence. Regular audits of detection results, including legitimate claims incorrectly flagged and fraudulent claims initially overlooked, can help carriers refine their systems while maintaining timely service for policyholders.