Insurers confront privacy risks in AI fraud fight

As AI transforms fraud detection, insurers face ethics and data protection challenges.

Artificial intelligence is transforming insurance fraud detection, pushing the industry from reactive systems to predictive, real-time prevention. Yet as insurers automate claims and cross-policy analysis, the need to balance innovation with privacy and ethical accountability is becoming increasingly urgent.

Roopa Malhotra, Head of Customer & Digital, APAC at Zurich Insurance, said the shift marks a major evolution in how insurers fight fraud. “[AI is] moving from static rules based systems to dynamic learning based models that can spot suspicious behaviors the moment it occurs,” she said.

By analysing large volumes of structured and unstructured data—from claims forms and invoices to images and geolocation metadata—AI can now “detect a pattern across these different data forms, which in the past, were invisible to human analysts or legacy systems,” she added.

The technology is also enabling cross-policy insights, with fraud indicators in one insurance line informing others. Malhotra noted that the industry is now moving toward “agentic AI,” which allows automation of “the entire end-to-end process.”

For Hoe Seng Chia, Senior Customer Success Manager at Shift Technology, AI adoption serves a broader societal purpose. “Our goal here is really for the better of the society. It is really to protect the honest policy holders and ensure that…claims which are legitimate…get paid out promptly and fairly,” he said. Keeping premiums affordable and maintaining a sustainable claims loss ratio, he added, are central to insurers’ long-term viability.

However, both experts agree that AI’s promise depends on trust. “AI and data privacy must evolve together. One can't advance responsibly without the other,” Malhotra said.

Chia added that insurers are building internal compliance teams to “prove and share openly that…our models do not include any bias,” underscoring the industry’s commitment to ethical, transparent AI adoption.