Artificial intelligence is expected to reshape insurance over time rather than trigger immediate disruption, but analysts at Moody’s warned emerging liability questions and growing data centre exposure are already creating new challenges for insurers and reinsurers.
Speaking during a Moody’s media briefing, analysts said AI could improve efficiency across underwriting, claims and distribution, particularly in retail insurance markets. However, uncertainty around casualty exposure, accumulation risk and AI-related litigation is creating new questions around how insurers assess and price evolving risks.
Salman Siddiqui, head of EMEA insurance credit ratings, said Moody’s has maintained stable outlooks across European P&C and life insurance markets, despite ongoing economic and geopolitical uncertainty.
“We have affirmed a stable outlook for both P&C or non-life in Europe, as well as the life outlook in Europe,” Siddiqui said. “There’s one positive outlook that’s on the Dutch life market and that is primarily driven by what we consider to be structural growth opportunities in the Dutch BPA or pension risk transfer market.”
Siddiqui also highlighted growing debate around how insurers maintain their role in the value chain as capital markets and insurance-linked securities continue to expand.
“At what point does the carrier become irrelevant?” he said. “How does a carrier maintain relevancy in the value chain?”
Brandan Holmes, senior credit officer, said Moody’s expects AI’s impact on insurers to emerge over time rather than through sudden disruption.
“Our key takeaways are that we think the impact will unfold gradually and on balance it would be sort of credit neutral to positive for the insurance sector,” Holmes said.
He said the main benefits are expected to come through operational efficiency and improved execution across underwriting and claims management.
“The main positives we see are the ability to extract efficiencies through the use of the technology and also improve execution across some functional areas such as claims management and underwriting,” Holmes said.
Holmes added that commercial and specialty insurance markets are likely to face slower disruption than mass-market retail insurance because of product complexity, long-dated liabilities and regulatory oversight.
“On the more specialty complex side of insurance, particularly on the commercial side, we definitely see AI more as an enabler to the current sort of market participants than a threat,” he said.
He added that the impact is likely to vary significantly across the sector depending on insurers’ scale, investment capability and execution.
Analysts also highlighted data centres as a growing source of opportunity and risk for insurers, with investment expected to grow significantly over the next five years.
Holmes pointed to Moody’s estimate of at least $3 trillion in global data centre investment over that period, creating significant demand for commercial insurance and reinsurance capacity. He said insurers face challenges around valuation, modelling and risk concentration because of the scale and geographic clustering of high-value infrastructure.
“One of the challenges still is managing or assessing and pricing this risk,” Holmes said.
Cyberattacks affecting physical infrastructure were also highlighted as a growing concern.
“If a cyberattack were to take out a cooling system, that could have a significant impact on the actual chips and the equipment in the data centre,” he said.
Matthew Harrison, senior director of strategy and innovation at Moody’s Insurance Solutions, said insurers are still trying to determine how AI-related liability risk will develop across casualty markets.
“It’s not clear that it’s generating whole new categories of risk, but as businesses become more and more AI enabled, it is changing the risks that they face,” he said.
Harrison said some comparisons are being drawn with “silent cyber”, although he noted the phrase “silent AI” was not one he particularly liked. The issue, he said, is less about whether coverage is being made explicit and more about the extent to which AI is becoming embedded across business activity.
He said the key uncertainty is whether liability ultimately concentrates around AI model developers or spreads across businesses using AI tools.
“If it’s around the usage and adoption, then it’s distributed at the user level and all of a sudden you have very large footprints which can expose many,” he said.
Harrison added that legal responsibility for AI-related errors remains uncertain as both adoption and legal theory continue to evolve.
