In a growing number of states, insurance companies are inspecting your home from aircraft and satellites, using AI to analyze what they find, and making coverage and pricing decisions based on that analysis — without setting foot on your property, without telling you an inspection is taking place, and without giving you a meaningful opportunity to contest what the algorithm sees.
If your policy is non-renewed, you may receive a letter citing a “property condition concern.” You will not receive the image that prompted the concern, the algorithm’s confidence score, or any way to know whether the AI correctly identified your roof or was analyzing the right address.
This is not a hypothetical scenario. NPR documented homeowners in Texas, California, Pennsylvania, and Florida who lost coverage or received non-renewal notices based on AI analysis of aerial images they had never seen. In at least one documented case, the AI was analyzing the wrong property.
How the System Works
The infrastructure enabling this practice is built on a set of specialized companies that aren’t well known to consumers.
EagleView Technologies, CoreLogic, Nearmap, and Cape Analytics are the primary vendors supplying AI-analyzed imagery to insurance carriers. These companies operate fleets of aircraft and maintain contracts for satellite imagery, producing regularly updated aerial views of virtually every property in the United States. The imagery is then processed through computer vision systems trained to identify attributes relevant to insurance risk:
- Roof condition: age, wear patterns, missing or damaged shingles, sagging sections
- Tree hazards: overhanging branches, dead trees within fall distance of the structure
- Property maintenance: yard conditions, debris, objects the algorithm classifies as risk indicators
- Defensible space (in wildfire regions): vegetation clearance around the structure
- Additions and structures: pools, trampolines, detached buildings, solar panels
Insurance carriers subscribe to these services as part of their underwriting workflow. When you apply for a policy, or when an existing policy comes up for renewal, the carrier may query these imagery databases to assess your property’s risk profile. The query happens without your knowledge.
Why Insurers Use It
From the insurer’s perspective, the appeal is straightforward. A human property inspection costs money, takes time, requires scheduling access, and happens at most once every few years. Aerial imagery is continuously updated, covers every insured property simultaneously, and requires no coordination with the homeowner.
SAS insurance analysts predicted last year that AI-driven underwriting would shift from an experimental add-on to a core operating system for insurance decisions. That prediction has materialized. Carriers are using aerial analysis not just for individual policy decisions but for portfolio-level risk assessment — identifying concentrations of properties they consider high-risk in specific geographies and adjusting their exposure accordingly.
For wildfire-prone regions in California, Oregon, and other western states, the pressure on insurers is acute enough that the practice has become near-universal. Carriers have left entire ZIP codes. The ones that remain often rely heavily on automated aerial analysis to differentiate properties within those zones.
The business case is clear. The privacy case has lagged far behind.
The Consent Problem
The central privacy issue is that this inspection happens without notification.
Your insurance policy — even if you have read it carefully — almost certainly does not say “we will conduct aerial and satellite inspections of your property using AI, and we may make coverage decisions based on the results.” The practice is permitted under the general language that allows insurers to assess risk and inspect insured properties, but the specific mechanism of AI-analyzed aerial imagery, deployed without any notice, sits in a gap between what that language was written to describe and what it is now being used to justify.
The homeowner in the NPR report who lost coverage had no idea an inspection had occurred. They received a non-renewal notice referencing property conditions. When they asked what had been observed, they were not provided with the imagery or the specific AI output. When they challenged the decision, they were told the determination was final.
This is the consent problem at its sharpest: a decision with significant financial consequences — the loss of home insurance, which may make a mortgage non-compliant and trigger lender requirements — made based on automated analysis of your property, with no notification, no right to see the evidence, and no clear appeal process.
The Error Problem
AI property analysis has a documented accuracy problem that the industry has been slow to address publicly.
Aerial imagery is taken at fixed intervals. An AI system analyzing images may be working from photos that are twelve or twenty-four months old. A roof that was repaired six months ago may still appear damaged in the imagery being analyzed. A tree that has been removed may still appear in the dataset.
Then there is the wrong-property problem. Addresses don’t always map cleanly to imagery. In dense urban areas, in neighborhoods with complex lot configurations, or simply through data errors in the address-to-coordinate mapping, the property an insurer’s system is analyzing may not be yours.
NPR documented at least one case where a homeowner’s coverage decision was based on imagery of an adjacent property. The homeowner could prove this — they had their own documentation of their roof’s condition — but the burden of proof was on them, not on the insurer.
The scale of this deployment means even a small error rate represents a large number of affected homeowners. If two percent of AI property assessments contain a meaningful error, and millions of policies are assessed this way annually, the absolute number of incorrect decisions is substantial.
The Regulatory Response (And Its Limits)
Several states have begun addressing this practice, though the regulatory landscape is fragmented and moves slowly.
California introduced legislation in 2025 requiring insurers to notify homeowners before using aerial data in coverage decisions and to provide a 30-day window to remediate identified issues before any coverage action takes effect. Whether this bill has passed into law and is being enforced depends on developments after this writing.
Louisiana passed a law in 2024 explicitly limiting the practice of aerial inspection without notice for certain property insurance decisions.
Colorado issued formal guidance in March 2026, directing insurers operating in the state to maintain transparency in how aerial imagery and algorithmic analysis contribute to underwriting decisions.
The National Association of Insurance Commissioners (NAIC) piloted a tool allowing state regulators to directly audit AI systems used in insurance underwriting — an attempt to create a verification mechanism for what the algorithms are actually doing.
These are real developments. They are also reactive, piecemeal, and mostly at the proposal or early-implementation stage. In most states, the practice of using AI-analyzed aerial imagery in insurance underwriting without homeowner notification continues without specific restriction.
What Rights You Actually Have
Your rights in this situation depend heavily on your state.
In states with explicit laws or guidance on aerial imagery in insurance:
- You may have the right to be notified before an aerial inspection is used in a coverage decision
- You may have the right to see the imagery or AI analysis that contributed to a decision
- You may have a remediation window to address identified issues before coverage action
In states without specific guidance (most states):
- You generally have the right to request the basis for a coverage decision under standard insurance regulations
- You have the right to file a complaint with your state’s insurance commissioner if you believe a decision was made in error
- You have the right to an independent appraisal in some coverage dispute contexts
The most actionable step if you receive a non-renewal or coverage change notice that cites property conditions you weren’t aware of: request, in writing, the specific evidence the insurer used to make that determination. Ask explicitly whether aerial or satellite imagery was used, and request copies of that imagery if it exists. Insurers are often reluctant to provide this, but requesting it creates a paper trail and may trigger disclosure requirements in some jurisdictions.
The Surveillance Parallel
The aerial imagery insurance situation is worth thinking about alongside other forms of automated data collection about your life — because the structural features are similar even when the surface looks very different.
In both cases:
- Data about you is collected without your participation or notification
- That data is analyzed by automated systems
- Decisions with real consequences are made based on the analysis
- The evidence underlying those decisions may be difficult to access or contest
- Your ability to opt out is limited or nonexistent
The data involved is different — your physical property versus your digital files and photos. The consequences are different — insurance pricing versus advertising targeting or AI model training. But the pattern of surveillance-based automated decision-making, operating without meaningful consent or transparency, applies to both.
The reason this matters for how you think about digital privacy is that the aerial imagery case makes the structure unusually visible. The homeowner who receives a non-renewal notice based on a photo they never saw can see clearly that something was done to them without their knowledge. The same structure exists when your photos are analyzed for training data, when your files are scanned for content classification, when your behavior is profiled for ad targeting — but those processes are less visible.
The aerial imagery case is useful precisely because it makes the abstract concrete: automated analysis of your private environment, affecting important decisions, conducted without your knowledge, with limited recourse.
What This Means for Personal Data Generally
You cannot opt out of satellites. Aerial imagery of your property will be collected regardless of your preferences, and the insurers who purchase access to that imagery will use it as long as it remains legal to do so. The protections that exist are regulatory, not individual.
Your digital data is different in one important way: you choose which services hold it, and different services have meaningfully different practices around analysis, use, and third-party access.
When you choose where to store personal photos and files, the relevant question isn’t only whether they’re secure from external breach — it’s also whether the service that holds them analyzes their contents, uses that analysis for its own purposes, or makes it available to third parties. A service that stores your files without analyzing them, without using them for advertising or AI training, and without selling access to them represents a different kind of privacy commitment than one that does.
The aerial imagery case is a reminder that “collected for one purpose” and “used for another” is not a hypothetical scenario. It’s an operational pattern across industries that will continue wherever data collection creates commercial value and consent requirements are weak or absent.