Insurers with stronger AI governance see higher returns

SAS has published research suggesting insurers with stronger AI governance are more likely to report higher returns from artificial intelligence. The findings are based on a global survey of 2,699 decision-makers conducted with IDC.

The study found that 54% of insurers have AI operating across several functions or deployed at company-wide scale. The clearest gains were reported by organisations classed as AI leaders, defined as those with stronger governance, data quality and auditability practices.

Among those leaders, 59% expect a strong or high return on investment from AI, while none of the laggards reported a comparable signal. Insurance was also the only industry in the survey where laggards showed no strong ROI expectation at all.

The divide was also evident in internal policy communication. According to the findings, 78% of leaders had communicated a responsible AI policy to every employee, compared with 3% of laggards.

Governance gap

The research argues that AI value depends less on the mere presence of the technology than on how organisations manage it. Companies with stronger controls, oversight and data standards were more likely to link AI use to operational results, while weaker performers were more likely to focus on cost-cutting.

Laggards most commonly named cost savings as the main goal for AI use. Leaders, by contrast, were more likely to prioritise operational efficiency, profit growth and revenue growth.

"The insurance industry demonstrates that, with AI, ROI follows trust. The insurers who are governing AI well are the ones seeing strong returns," said Dr Chris Marshall, VP for Enterprise AI Strategy at IDC.

Franklin Manchester, Global Strategic Insurance Advisor at SAS, said the findings suggest some insurers still approach AI too narrowly.

"Whenever I hear executives say they're investing in AI in order to reduce staff, it tells me the organisation has no strategy, because cutting costs is the least imaginative use for AI. Keep your people, teach them how to use the tools safely, and pursue goals like improved customer experience, expanded market share and stronger resiliency," said Franklin Manchester, Global Strategic Insurance Advisor at SAS.

Rising spend

Insurers are also planning to spend more on AI. The survey found that 76% expect to increase investment over the next year, while 84% expect a positive return. More than six in 10 expect that return to be 50% or more.

Those expectations are tied more closely to customer-facing improvements than to workforce reduction, according to the data. Insurance is the first industry in the survey to close the gap between trust and capability, with perceived trust in AI now slightly higher than measured capability.

That does not mean insurers accept AI outputs without challenge. Across industries, nearly 80% of respondents said their organisations override more than 10% of decisions made by AI agents, and nearly a third said they do so more than 30% of the time.

The most common reason for overriding an AI decision was a lack of explanation, cited by 34.5% of respondents. A combined 40% pointed to bias or unfairness, factual errors, or decisions they believed were unsafe, with 8% specifically citing safety concerns.

Manchester said that figure should not be dismissed given the volume of decisions AI systems can influence.

"The concerns over safety are a red flag. Eight per cent may sound like a small number, but when you consider the speed at which AI generates output, and the thousands of decisions made by insurers every day - never mind bankers, healthcare providers and other industries - that small percentage translates into a very large number of potentially unsafe actions," he said.

European picture

The survey also pointed to a mixed picture in Europe. Insurers in the region improved their Trustworthiness Index score from 41.9 to 60.6 year on year, indicating stronger governance, data and oversight practices.

Even so, the result remained below the global insurance benchmark of 63.2. Perceived trust in Europe edged down from 70.2 to 69.1, while the AI Impact Index, which measures realised business value, stood at 53.0 against a global insurance figure of 55.9.

The figures suggest European insurers have improved the structures around AI more quickly than they have turned those improvements into measurable commercial results. This leaves scope for the region's insurers to convert stronger governance into greater confidence and better outcomes.

Marshall said the next step is to build adoption and confidence together rather than treat them separately.

"The path forward is to scale AI deployment and user trust in tandem so that reliability and confidence can grow together," he said.