Nearly $1 Trillion in California Homes Labeled "Low Risk" Despite Elevated Wildfire Danger

One year after LA wildfires, analysis of 11.8 million properties with AI models reveals elevated risk that federal maps miss.

SAN FRANCISCO, Calif., Jan. 8, 2026 /PRNewswire/ - One year after the Palisades and Eaton fires devastated Los Angeles County, a new analysis of 11.8 million California residential properties finds that approximately $1 trillion in residential property value is classified as low wildfire risk despite elevated fire danger across California, including parts of Los Angeles County affected by the January 2025 fires.

The statewide analysis examined property-level wildfire risk for every residential structure in California using ZestyAI's Z-FIRE™ model and compared the results with FEMA's National Risk Index, which assesses wildfire risk at the census-tract level. The findings reveal that 1.2 million California homes are designated as 'Very Low,' 'Relatively Low,' or 'No Rating' for wildfire risk under federal assessments, despite AI models identifying elevated wildfire danger at the property level—a classification gap that persists across the state.

The same classification gap is visible within the LA fire zones themselves. In the areas impacted by the Palisades and Eaton fires, the analysis identified 3,045 properties, worth an estimated $2.4 billion, that remain designated as low or nonexistent wildfire risk under federal classifications, despite property-level AI models flagging elevated wildfire danger in these neighborhoods before the fires occurred.

Palisades fire zone (Pacific Palisades, Malibu):

• 1,430 properties flagged as elevated wildfire risk by AI models but classified as low or no risk under federal wildfire assessments
• $1.14 billion in estimated residential property value at risk
• 229 homes rated very-high to extreme wildfire risk despite low-risk federal classifications

Eaton fire zone (Altadena, Pasadena, Sierra Madre, La Cañada Flintridge):

• 1,615 properties flagged as elevated wildfire risk by AI models but classified as low or no risk under federal wildfire assessments
• $1.29 billion in property value at risk. All located in census tracts FEMA rates "Very Low," "Relatively Low," or "No Rating"

Combined, these fire-devastated communities illustrate a pattern that persists statewide: wildfire risk is best understood by incorporating property-level conditions, not just regional averages — a distinction federal maps are not designed to capture.

"Wildfire risk today is driven by property-level conditions, not regional averages," said Kumar Dhuvur, co-founder and Chief Product Officer at ZestyAI. "Federal maps operate at a scale that can't see vegetation, defensible space, or building characteristics — the very factors that determine loss. That's why the same blind spots that existed before the LA fires still exist across California."

Across California:

• 1.2 million properties rate as high wildfire risk on AI models but low or no risk on FEMA maps
• $940 billion in property value sits in these blind spots, based on a conservative median home price of $800,000
• Only 37.5% of California properties carry any FEMA wildfire rating; 62.5% have no federal wildfire assessment at all
• 300,859 blind-spot homes were built before 1980, predating modern fire-resistant building standards
• Significant concentrations appear in San Diego, San Bernardino, and Riverside counties, with additional clusters across the Inland Empire, Sierra Nevada foothills, and North Coast

Unlike static, census-level federal assessments, property-level wildfire risk evolves over time as vegetation regrows, development expands, and mitigation measures are added or removed. ZestyAI's analysis tracked how both hazard exposure and home destruction risk changed between 2022 and 2025 across millions of California properties — revealing meaningful shifts that federal wildfire ratings do not reflect.

Among high-hazard risk properties (score ≥ 7):

• 41.3% reduced home-destruction risk through visible mitigation
• 34.2% experienced increased wildfire hazard in their surrounding environment

Among extreme-hazard risk properties (score ≥ 9):

• 42.5% improved their home destruction risk through mitigation
• 52.0% saw regional wildfire hazard worsen, with scores increasing an average of 1.3 points. To put that in perspective: a single-point increase represents a 51% jump in annual wildfire probability—meaning environmental conditions deteriorated faster than homeowners could mitigate.

Over the same period, FEMA wildfire ratings showed little change, unable to reflect mitigation, vegetation regrowth, or shifting conditions on the ground.

Ends

Methodology

The analysis compares ZestyAI's Z-FIRE™ AI wildfire risk model with FEMA's National Risk Index for 11,808,251 California properties with Z-FIRE scores as of March 2025. Z-FIRE uses machine learning and annual high-resolution satellite imagery to evaluate vegetation, terrain, building characteristics, climate factors, and proximity to historical wildfires, producing 1–10 scores for wildfire hazard (L1) and potential home destruction (L2).

FEMA's National Risk Index evaluates wildfire risk at the census-tract level based on historical fire perimeters, exposure, social vulnerability, and community resilience. Property value exposure is estimated using California's median home price of $800,000. Year-over-year analysis covers 9,266,054 properties from 2022 to 2025.

About ZestyAI
ZestyAI is the Risk and Decision Intelligence Platform for the insurance industry. Trusted by Property and Casualty carriers and regulators across the United States, ZestyAI helps insurers make better decisions faster and with greater confidence. The platform unifies property-level data, predictive AI models, and Agentic AI automation to transform how insurers see, price, and manage risk. Through machine learning, computer vision, and regulatory-grade transparency, ZestyAI delivers precision and performance across underwriting, rating, reinsurance, and regulatory workflows.

Validated by climate science and historical loss data, ZestyAI's models cover major perils including wildfire, severe convective storm, and non-weather water. From improving pricing accuracy to strengthening reinsurance outcomes, ZestyAI brings trusted AI to every insurance decision—helping the industry operate with speed, accuracy, and resilience.

SOURCE ZestyAI