Sharper risk pricing through artificial intelligence could make some risks harder or more expensive to insure, according to APRA member Suzanne Smith (pictured). In the same speech, she said brokers and customers would increasingly use AI to scrutinise insurers’ products and decisions.
Smith oversees general, life and private health insurance at the Australian Prudential Regulation Authority. She told the Insurance Council of Australia’s annual conference in Sydney on Friday that more precise risk assessment could improve underwriting and pricing. Greater segmentation, though, could also affect how broad, accessible and affordable insurance pools remain.
She said the shift raised questions about where the limits of insurability lie. It also raised questions about who should do the work of reducing underlying risk: insurers, regulators, governments or communities. APRA’s own expectation was unchanged. Insurers must understand and properly price the risks they take on, and stay financially able to keep their promises to policyholders.
Smith said AI would change how insurance is distributed, compared and chosen. Policyholders and intermediaries would increasingly use it to compare cover, prepare claims and contest decisions. That could improve customer understanding and competition, she said, but it would also put more weight on products being transparent and easy for those tools to find and interpret.
She warned that the same trend could add to insurers’ costs through unreliable AI-generated advice, larger volumes of claims material, AI-generated legal arguments and new forms of fraud. She added that AI could erode brand loyalty and increase anti-selection risk.
The speech follows APRA’s April letter to industry, which found that governance, risk management, assurance and operational resilience were not keeping pace with AI adoption. Smith said boards did not need to understand every technical detail. They did need to know where AI is used, what its material risks are and who is accountable for it.
“AI can support judgement; it cannot replace accountability,” said Smith.
She said advanced technology would become a weaker differentiator as it became easier to acquire. Insurers would instead be set apart by financial resilience, governance, reliable operations, skilled staff and the ability to adapt. She urged insurers to stop treating technology, resilience and customer outcomes as separate agendas, and to check whether investment in each was strengthening the others.
The two strands of the speech pull in different directions for intermediaries.
AI comparison tools could make it easier for brokers to test products against each other, and to push back on claims decisions when wordings are clear enough for those tools to read. Insurance Business reported in September that insurers’ uneven confidence in their own AI use is already a live issue for brokers.
The segmentation warning points the other way. If finer pricing moves more clients outside standard pools, those placements become harder and more dependent on broker skill. Clients with high-hazard properties or difficult occupations are the most likely to be affected. The point comes as ICA chief executive Andrew Hall has warned that some Australian homes will never be affordably insurable.
Smith did not announce new rules, and the speech sets no compliance deadline. Even so, APRA now openly accepts two things: sharper pricing may narrow insurance pools, and intermediaries will use AI to test insurers’ products and decisions. In that market, brokers are likely to spend more time on the risks that algorithms price out, and on clients who want those algorithms’ decisions explained.
