The AI Insurance Trap: California’s Property Market Death Spiral

If you live in California and wonder why your homeowner’s insurance bill looks like a ransom note, congratulations: you have discovered Proposition 103. The political class has a more convenient explanation—artificial intelligence is discriminating against your bungalow—but the algorithmic villain is a distraction from the man behind the curtain, who has been there since 1988.

Since 2021, the California FAIR Plan—the state’s insurer of last resort, created as a market of necessity, not a market of choice—has grown 139 percent, to nearly 574,000 policies with total insured exposure approaching $600 billion. That is not a market in transition. That is a market in full retreat, and the retreating army left because Sacramento made it mathematically impossible to stay.

State Farm, California’s largest property insurer, has not written a new homeowner policy in the state since May 2023. In 2024 it announced the non-renewal of 72,000 policies in its highest-risk ZIP codes. Its own disclosures show it paid $1.26 for every premium dollar collected over nine years—more than $5 billion in cumulative losses. Allstate exited new applications in 2023. These are not arbitrary corporate decisions. They are actuarial triage performed by companies that could no longer reconcile their risk models with the premiums Sacramento permitted them to charge.

The culprit is Proposition 103, the 1988 ballot initiative that subjected all property and casualty insurance rate increases to prior approval by an elected Insurance Commissioner. The Commissioner has both the power and the political incentive to deny increases regardless of actuarial merit. Reinsurance markets reprice annually using real-time catastrophe models. California’s approval process can take years. An insurer that cannot charge a rate commensurate with its wildfire risk exposure has exactly two choices: exit or bleed. The International Center for Law and Economics has documented that California ranks worst in the nation for rate suppression. The market responded accordingly.

Proposition 103 was deliberately drafted to resist amendment—requiring a two-thirds supermajority of both legislative chambers or a direct popular vote to modify—which has made Sacramento structurally incapable of reforming its own creation. The mechanism for eliminating a market without technically banning it is by now familiar: cap rates through prior approval; expose carriers to expanded bad-faith litigation when they try to modernize their risk tools; politicize every rate hearing through the Consumer Watchdog intervenor system whose founder wrote Proposition 103 in the first place; then redirect displaced homeowners to a government backstop that was never designed to handle majority-market exposure. The FAIR Plan now carries nearly six percent of the state’s total property insurance market. In some wildfire-prone counties, it covers more than fifty percent.

California’s legislature responded to the crisis by passing SB-1120 in 2024, restricting AI use in claims decisions, and the National Association of Insurance Commissioners issued AI guidance in 2023. These are sensible consumer protections. They are also irrelevant to the structural problem: State Farm’s exit predates every high-profile AI lawsuit in the homeowner space. The FAIR Plan’s exponential growth began before satellite-imagery litigation became a legislative preoccupation. Regulating the conduct of insurers who have already left the market is the equivalent of posting a speed limit on a road nobody drives anymore.

The FAIR Plan’s own fragility makes the stakes concrete. Following the January 2025 wildfires, the FAIR Plan paid out $2.7 billion in claims and levied a $1 billion emergency assessment on member insurers to avoid insolvency. Its president has publicly stated the organization is praying for a mild fire season. An institution managing $600 billion in exposure on prayer is not a safety net. It is a tightrope over a canyon.

The solution is structural and uncomplicated: deregulate rate-setting. Let insurers price wildfire risk using current catastrophe models. Let premiums reflect the actual risk profile of the property, not the political preferences of an elected Commissioner in an election year. Florida, Texas, and Oklahoma all operate private property insurance markets in high-catastrophe environments. They function because they permit risk-based pricing. California’s does not, and every year of delay is another year of carriers making the rational decision to leave.

California has approximately one fire season to decide whether it wants to govern its insurance market based on arithmetic or ideology. The arithmetic is patient. The wildfire calendar is not.