SAS study: Insurers see AI returns when governance and trust come first

Leaders and laggards split on ROI

The report, titled The New Economics of Trust, divides organizations into trustworthy AI "leaders", those with the strongest governance, data quality and auditability practices, and "laggards". According to SAS, insurance is the industry most strategically committed to AI and shows the clearest gap between the two groups.

Over half of insurers (54%) have AI operational across several functions or scaled companywide. Among insurance AI leaders, 59% expect strong or high AI ROI, compared with 0% of laggards. SAS says insurance is the only industry in which laggards produced no strong ROI signal at all.

Transparency inside the organization shows a similar divide. In the report, 78% of leaders have communicated a responsible AI policy to every employee, versus just 3% of laggards.

Cost cutting ranks among the least effective AI goals

The report says deploying AI mainly to cut costs is among its least effective uses, which it describes as one of its "most durable findings". The most common goal named by laggards is cost savings. Leaders instead prioritize operational efficiency, increasing profits and revenue growth.

Franklin Manchester, Global Strategic Insurance Advisor at SAS, was blunt about it. "Cutting costs is the least imaginative use for AI," he said, adding: "Keep your people, teach them how to use the tools safely, and pursue goals like improved customer experience, expanded market share and stronger resiliency."

Investment plans point the same way. Over 3 in 4 insurers (76%) expect to increase their AI investment in the next year. A large majority (84%) expect a positive return, and over 6 in 10 anticipate that ROI to be 50% or more. SAS says those gains are expected from customer-facing outcomes rather than cost reduction.

Trust has caught up with capability, but overrides remain

Insurers are the first and only industry in which perceived trust in AI exceeds measured capability, by less than 1%, according to the report.

Concerns persist, though. Across industries, nearly 80% of respondents say their organizations override more than 10% of decisions by AI agents, and nearly a third do so more than 30% of the time. The most frequent reason for overriding AI was lack of explanation for its decision, at 34.5%. A combined 40% cited bias or unfairness, factual incorrectness or decisions they felt were unsafe, with unsafe decisions accounting for 8%.

Manchester called the safety concern "a red flag". He argued that given the speed of AI output and the thousands of decisions insurers make every day, a small percentage can translate into a very large number of potentially unsafe actions.

What it means for the industry

The report is based on a survey conducted in April 2026 of 2,699 decision-makers across 28 countries and five regions, with a focus on banking, insurance, life sciences and the public sector.

Dr. Chris Marshall, VP for Enterprise AI Strategy at IDC, said the insurance industry shows that "ROI follows trust". He added that the path forward is to scale AI deployment and user trust in tandem so that reliability and confidence can grow together.

For insurers, the findings suggest that governance and explainability are tied to financial results, not only compliance. SAS, KPMG and Clal Insurance will discuss the topic in a live webinar, The State of AI Trustworthiness in Insurance, at 9 a.m. ET on Oct. 15.