Recent data on insurance fraud is hard to come by. Estimates suggest that 10% of claims are either fabricated or inflated. However, some industry stakeholders are raising concerns that artificial intelligence (AI) could drive a new wave of fraudulent claims.
“The potential disruption wrought by AI-driven insurance fraud is a pressing concern,” said Leah Hewish (pictured), partner at Clyde & Co, in an interview with Insurance Business. “This new battleground has infiltrated claims handling across business lines.”
The Sydney-based insurance law specialist said these cases are requiring additional attention from claims teams who are already under pressure to comply with the regulator’s time-sensitive claim handling rules.
Hewish said the potential use of fraudulent supporting evidence in an insurance claim is an “urgent challenge” for insurers.
“It dramatically upsets the long-established premise that insurance policies are based on mutual obligations of good faith,” she said.
Hewish said this industry-wide issue is best addressed through partnerships and training. She referred to existing industry collaborations including the Insurance Council of Australia’s (ICA’s) counter-fraud and scams operation.
“Awareness of the proliferation of AI-driven insurance fraud is crucial, including internal training and industry collaboration relating to trends, new technology and uncovered scams,” she said.
Ironically, AI is also a powerful partner as insurance firms strengthen their defences against AI driven fraud.
According to industry reports, AI is helping insurers develop new fraud detection models, particularly by combining and analysing a wide range of data from customers.
“Combining data from various modalities, such as text, images, audio, and video, could help identify patterns and anomalies and enhance the investigative process by reducing false positives, increasing detection rates of fraudulent claims, and saving on costs associated with fraud investigations,” said a recent Deloitte report.
Another report from Hicron Software said machine learning has become a “game-changing tool in the insurance sector, especially for tackling fraud.” Algorithms, said the report, efficiently detect unusual patterns that could signal fraudulent behaviour.
Natural Language Processing (NLP) is an example of an AI technology used in insurance fraud detection.
“NLP algorithms analyse unstructured text data – such as claims descriptions, customer communications, and social media interactions – to extract valuable insights about potential fraudulent behaviour,” said a blog on tech company Inaza’s website.
The blog said “AI is revolutionizing the way insurance fraud is detected and prevented.”
Hewish warned that detection is one thing, proving insurance fraud is another.
She said that denying fraudulent insurance claims and prosecuting offenders is not uncommon. However, a major challenge for Australian insurers in court, said Hewish, is the requirement for strong and compelling evidence in support of an allegation of fraud.
The Insurance Council of Australia (ICA) raised the issue of AI driven insurance fraud in a submission to the government’s 2024 discussion paper looking at the adequacy of AI regulation.
The ICA mentioned AI augmented photos that are used to inflate claims.
“Similarly, AI could be used for voice spoofing, email phishing, or synthetic identity fraud,” the ICA said. “Commercial sensitive data such as proprietary information uploaded to a public AI model may then be shared elsewhere by the model.”
The peak body also said AI can significantly streamline operational efficiency and improve customer service. By reducing costs linked to claims processing and risk evaluation, the ICA said AI can also make insurance more affordable.
A major recent case of insurance fraud involved the National Disability Insurance Agency (NDIA). By 2024 five NSW residents had been sentenced to prison for their roles in defrauding the National Disability Insurance Scheme (NDIS) of $5.8 million.
During the investigation that started in 2020, the release said police seized 8 kg of gold bullion, $600,000 in cash from multiple residential properties, $635,000 in cryptocurrency, jewellery and three vehicles, including a BMW, an Audi and a Porsche.
How do you see AI and insurance fraud? What’s the biggest challenge for brokers and insurers? Please tell us your thoughts below.
