Amid Climate Crisis, Insurers’ Increased Use of AI Raises Concern For Policyholders

William May’s home in Pacific Palisades was destroyed in the L.A. wildfires in January 2025. He’s still haunted by the memory of the “fireball burning everything in its path” on that hellish day. And all he wants to do is rebuild his beautiful home, where the retired pediatrician lived with his wife.

Since then, he’s been fighting with State Farm, his property insurer, to get the money he said he needs to rebuild his home. Back in 2017, when he bought the two-story home, he said it was valued at $1.7 million. But the insurer gave him an estimate of only $1.35 million after the fire, and May said he’s driven himself into debt trying to rebuild the couple’s home while they wait for State Farm to reassess their claim. Property values in the neighborhood have increased 50% from $2.1 million on average in December 2017 to $3.076 million in December 2025, according to Zillow’s Home Values Index.

“How can it be worth less now than it was when it was new?”

May blames State Farm’s use of an AI-powered software called Xactimate, which the insurer employs to estimate property repair, rebuilding and cleaning costs. “They use this reductive method. It’s a phony way of calculating every screw, every bolt, and coming up with a profit for State Farm by undervaluing the house.”

May considers himself lucky, since he has the resources to rebuild, noting that many of his neighbors can’t afford to do that and face similar problems with their insurers. He also blames Verisk, the data analytics company that makes Xactimate.

“I’m pretty sure these companies make these programs just to sell to insurance companies so that they can lowball people because the insurers are interested in squeezing people for profit,” he said.

A spokesperson for State Farm told Capital & Main: “State Farm remains committed to helping our customers throughout the entire recovery process and paying them all benefits available under their policies. So far, we’ve issued over $5 billion in payments to families whose homes, cars, and property were damaged or destroyed in the fires. We encourage any customer with questions or concerns to reach out to us.”

A spokesperson for Verisk said that ”Xactimate’s AI capabilities support tasks such as summarizing information or labeling photos and always operate under human review and control.” She added that Xactimate “does not generate repair costs using AI, and AI does not determine the price of materials, labor or reconstruction. Xactware’s construction cost database is market-based, transparent and rooted in human-validated data. It is intended as a flexible benchmark that users can adjust for specific jobs and local conditions.”

State Farm, along with other major property insurers, is increasingly turning to artificial intelligence to increase efficiency and improve risk modeling. The insurer posted a net income of $5.3 billion in 2024, a turnaround from a $6.3 billion loss in 2023. In late 2024, the company’s former vice president of innovation and venture capital, Haden Kirkpatrick, said in an interview that AI and other emerging technologies will help the industry “better predict and prevent losses.” 

As the insurance industry grapples with the climate crisis — more extreme weather events destroying homes, leading to greater losses and skyrocketing premiums — AI has been touted as a game-changing asset. By analyzing vast datasets, the technology has the potential to predict and manage risk more accurately, improving underwriting efficiency and even enabling insurers to offer coverage in areas that otherwise would be considered uninsurable due to climate volatility.

Yet AI’s performance in recent years has been criticized for inaccurate predictions when it comes to climate change, algorithmic bias, privacy concerns, lack of transparency and incorrect outputs such as “hallucinations.” Industry watchdogs have raised concerns that insurers could rely on the technology to make quick decisions in the name of cost efficiency with complicated claims that require human analysis.

From California to Alabama to Illinois, policyholders and prosecutors have filed lawsuits claiming that property insurers’ use of AI has allowed them to underpay claims, discriminate against nonwhite customers and drop coverage altogether. The class-action suits have focused on what’s called AI-washing – when the technology is misapplied to manage risk in a way that hurts policyholders.

In the wake of complaints by homeowners like May, Los Angeles County recently announced a probe into State Farm’s use of AI tools that allegedly delayed or denied claims. The county’s counsel sent a letter to the insurer in November seeking documents related to the L.A. wildfires: “Any and all documents, including but not limited to memoranda, bulletins, manuals, training materials, policy statements, guidelines, or directives that reflect, describe, or constitute State Farm’s use of Artificial Intelligence (AI) tools in the claims review process.”

State Farm announced in March 2024 that it would not renew about 72,000 California property insurance policies through 2025, citing wildfire risks and associated costs.

The insurer said in an online update on its California recovery response: “Recovery, following a catastrophe, doesn’t move in a straight line.” It added that “many families are engaged in the process of rebuilding and recovering from the devastation” and that “many families continue to navigate through parts of the claims process, with State Farm trying to address the needs of their unique circumstances.”

The insurance industry touts AI as a tool to help it with risk modeling as climate change increases in severity. It’s also optimistic that the technology will help it curb losses and improve its bottom line. In a policy paper by Bain & Co., consultants said they anticipate that generative AI will lead to a “30% to 50% decrease in total leakage — the difference between what is paid vs. what is owed per the contract, which occurs when adjusters deviate from policy guidelines or when supply chain problems cause unanticipated costs.” 

In a recent white paper by CAPE Analytics, which specializes in AI-powered property risk intelligence it sells to insurers, the company noted several reasons why the technology is needed to sift through a “mountain of contradictory data,” and noted that it can help insurers avoid providing too much coverage at low rates. Without AI, “The consequences of operating with raw data or drawing the wrong conclusions from it can lead to excessive exposure when quotes are too low and premium loss when they’re unnecessarily high.”

To insurance professionals and advocates for policyholders, that raises concerns that insurers will rely on the technology to make hasty decisions in often-complicated claims processes. 

“For example, there is the potential for AI systems to make decisions based on incomplete or biased data, leading to unfair treatment of policyholders,” noted Chip Merlin, a Florida lawyer who has represented policyholders. 

He cited a 2022 class-action suit brought by homeowners in Illinois who claimed that State Farm’s use of algorithms in its claims-processing methods disproportionately impacts Black policyholders, causing delays in repairs and the payment of benefits. The case is pending and the insurer insists that its practices do not violate federal law.

The biggest factors impacting the affordability and availability of insurance are climate change and technology like AI, said Amy Bach, the executive director of United Policyholders, an advocacy group. 

“Now they’re no longer willing to insure many people, and a lot of that is because of data and the use of AI in predictive analytics, as well as aerial surveillance. When people ask me, ‘What benefits are consumers getting from AI?’ I’m like, in the insurance context, none.”

Monica Palmeira, associate director of economic equity at the nonprofit Greenling Institute, said that AI could be used to increase bluelining — a modern version of redlining, describing a practice in which financial institutions and insurers withdraw from poor neighborhoods or dramatically hike rates in areas considered high risk for climate change.

When Palmeira and her colleagues started studying communities considered vulnerable to climate impacts, “We saw this pattern of the same communities that were excluded from financial services in the past continue to come up for exclusion today.” 

Insurance, she said, is one of the “first ways that communities experience that withdrawal of financial services and now it’s starting to be whole areas that can’t get insurance and that means they can’t get mortgages. So this contagion starts to happen.”

To address such concerns, states are taking action to protect consumers. One of the common themes of such measures is greater transparency — requiring that consumers be informed when AI is used in decision making, that companies maintain guidelines for the responsible use of AI, and make their policies and procedures for the use of AI publicly accessible. Those requirements are included in a guidance from the National Association of Insurance Commissioners, which provides a framework for the responsible use of AI.

At the same time, some lawmakers are pushing for human review in decision making by insurers. State Rep. Hillary Cassel, a Florida lawmaker, recently sponsored legislation that seeks to ensure humans make the ultimate decisions when it comes to insurance claims. 

“I think insurance companies should be allowed to use AI as a tool because premiums are very high across the country, especially here in Florida, and if insurance companies can use it to aggregate their resources and pass that savings on to consumers by using those types of tools,” she told Capital & Main. “But we also know that AI can be used for nefarious reasons, and I thought it was really important that in the space of dealing with denials that computers don’t always get it right.” 

The state-level action gives hope to Palmeira that consumers will start to “see the level of review and transparency that they really deserve.” She noted that the National Association of Insurance Commissioners experienced protests by consumers at a meeting, shortly before it announced its guidance on AI. “The way the insurance industry has been able to exert its power over insurance regulation has been so dominant for so long that we’re finally maybe starting to see a small shift in the tide.”

Palmeira acknowledged that climate change might make certain parts of the country uninsurable and that AI can be utilized in beneficial ways by insurers to improve their risk modeling and predictive analytics. 

“But it shouldn’t be black box models — without a human check on their decisions or without working with communities to make informed decisions about their livelihoods and wellbeing.”

One potential tool is parametric insurance — which utilizes data such as satellite imagery, IoT (Internet of Things) sensors that detect changes in the environment and weather feeds to trigger automated payments to policyholders when specific weather conditions are met for a particular home. It can be a “really useful tool to make sure folks have some kind of baseline coverage in a way that can be deployed very efficiently,” Palmeira said.

In addition, she suggests more community-based measures — such as a local government purchasing an insurance policy for an entire neighborhood or “where there needs to be more sensitive and serious conversations about relocation in very defined areas.”

Copyright 2026 Capital & Main