Event data recorders as evidence in claims and fraud cases

A car's event data recorder (EDR) keeps a few seconds of numbers around a serious impact: how fast the car was going, whether the brake and throttle were in use, whether the driver's belt was buckled. Investigators read those numbers beside the dents and the driver's story, and when the three disagree the claim gets a hard look. The data is the owner's, so an insurer generally needs your consent or a court order to read it, but it is evidence that can clear an honest driver as easily as it can undo a false account.

This article is about the evidence side: how a download is done, how adjusters and special investigators use the result, where the federal rules and real cases show its limits, and what to do if your own insurer asks for it. The basics of what the unit stores and who owns it are in event data recorders: what your car records and who can access it; this piece assumes you have read that or will.

What an EDR can and cannot show

The federal rule, 49 CFR Part 563, sets a floor for what a covered recorder must capture. It applies to cars, light trucks and vans made on or after September 1, 2012 that are equipped with an EDR and weigh 8,500 pounds or less (with an unloaded weight of 5,500 pounds or less), and it does not require a manufacturer to install one (eCFR, 49 CFR Part 563). When a unit is installed and a trigger event happens, it must record items including the change in speed during the crash, vehicle speed, accelerator or throttle position, whether the service brake was on, whether the driver's belt was buckled one second before, and whether the airbag warning lamp was lit.

Two details of the regulation matter more in a dispute than the list itself.

• The pre-crash window is short. As the regulation reads in November 2026, the baseline for a vehicle built before September 1, 2031 is only the 5 seconds before the event, sampled twice a second. A phase-in that starts with vehicles built from September 1, 2028 requires 25 percent of each manufacturer's production to carry the longer recording of 20 seconds sampled ten times a second, rising to 50 and 75 percent in later model years. So a typical car on the road today offers roughly ten speed readings leading up to the impact, not a movie of the drive.
• Accuracy is specified, and it is loose in places. The same rule's data-format table allows recorded speed to be off by 1 km/h (about 0.6 mph), accelerator position by 5 percent, and the crash change in speed by 10 percent. The speed field is called vehicle-indicated speed, meaning what the car's own sensors reported, which is not the same as true speed over the ground.

What the unit does not hold is just as important. It has no audio, no video, no map, and, under the rule's required owner's-manual statement, it records nothing during ordinary driving and no personal data such as name or crash location (49 CFR 563.11). Anyone who reads a report is looking at a handful of channels over a few seconds. The skill is in what you compare them with.

Who may read it: consent, court orders and a few state rules

The Driver Privacy Act of 2015 says data retained by an EDR is the property of the vehicle's owner or, for a leased vehicle, the lessee. It bars anyone else from accessing the data unless a court or other judicial or administrative authority authorizes retrieval, the owner or lessee consents in writing, electronically or by recorded audio, the retrieval is part of a federal safety investigation, it is used to support the emergency medical response to a crash, or it is used for traffic safety research with personal information and the VIN withheld (note to 49 U.S.C. 30101). An insurance company is not on that list, which means its routes are your consent or a court order.

Some states add rules aimed squarely at insurers. Two examples show how different the answers can be:

• Oregon. The state's EDR statutes say the recorder's data is exclusively owned by the vehicle's owner and, apart from narrow exceptions for court-ordered law enforcement retrieval, emergency response, research and repair, may not be retrieved by anyone else without the owner's written consent. An insurer may not condition payment or settlement of the owner's claim on consent, and may not require consent as a condition of issuing the policy. Data does not become the insurer's property just because the insurer takes over a wrecked car. An insurer can ask a circuit court for an order, but it must show that the owner has a policy with it, that the data is necessary to reconstruct the accident and determine its obligations, and that an accurate and timely determination cannot be made without it; the owner must be served at least 30 days before the hearing (ORS 105.928 to 105.938).
• North Dakota. An insurer may not require the owner's consent to access EDR data as a condition of insurability, and may not use data retrieved with consent, before or after an accident, to assess rates (N.D.C.C. 51-07-28).

Many states have EDR statutes with no insurer-specific language, and some say nothing on the point. In those states the question of whether a refusal to consent can breach your policy is decided under the policy's own words. The standard personal auto form published by Insurance Services Office (the PP 00 01 edition dated 09 18, which many insurers base their policies on) requires a person seeking coverage to cooperate in the investigation, submit to recorded statements and examination under oath, authorize the insurer to obtain other pertinent records, and permit the insurer to inspect the damaged property before repair or disposal (ISO PP 00 01 09 18, as filed with the Virginia SCC). Whether a download falls inside that language is a legal question that depends on your state and your policy. Before refusing a request outright, call the state insurance department or a licensed attorney, and see the cooperation clause: records, interviews and exams under oath for the wider duty.

How a download works

A technician connects a retrieval tool to the car and produces a report file. The best-known tool in the United States is the Bosch Crash Data Retrieval (CDR) system, which Bosch says has been used since 2000 by law enforcement, crash researchers, manufacturers and government agencies. Its standard kit reads data through the car's diagnostic port, called the diagnostic link connector, and needs a Windows computer (Bosch CDR). If a severe crash has damaged the car's electrical system, the port may not work, and the technician connects directly to the module that stores the data using separate cables, known as direct-to-module kits.

Federal rules are the reason this can be done at all. Part 563.12 requires each manufacturer of a vehicle with an EDR to make sure, by licensing agreement or other means, that a retrieval tool is commercially available no later than 90 days after the vehicle's first sale. Tool coverage varies by make, model year and software version, so a shop has to check the specific vehicle.

A careful report identifies the vehicle by VIN, the tool and software version, the date of retrieval and the technician. The rule also requires the recorder to count ignition cycles at the crash and at the time of download, so an analyst can see how many times the car was started in between. A large gap suggests the car kept being driven, which raises the question of whether a later, smaller event overwrote an earlier one.

How adjusters and investigators use the result

Most claims never involve a download. It is a tool for the claim that does not add up: a severe loss with a thin explanation, a late report, an injury claim out of proportion to the damage, or a "hit-and-run" with no witness (hit-and-run and phantom vehicle claims: the physical contact rule). The common uses are these.

Does the story fit the event?

Investigators compare the recorded speed, braking, steering and belt status with the statement the driver gave. A driver who says they were stopped at a light and were hit from behind should show a speed near zero with the brake applied. A driver who says the car was parked overnight when it was struck should show nothing at all, or at least no event at the claimed time.

A forensic-engineering article in Claims Journal (November 2023) describes the reasoning when no event is recovered: the unit records only when the car is powered, and only when the impact crosses its trigger threshold. If the reported loss says the car was off, no stored event is expected. But if the damage is consistent with a fairly severe collision and the recorder holds nothing, the authors say the car was most likely off at the time and the reported sequence of events is probably untrue (Claims Journal). The same article notes that the opposite is possible: a minor impact may simply fall below the threshold, which is why an empty record is a reason to keep investigating and not a conclusion.

Are there too many events, or the wrong ones?

A staged or inflated loss can leave the wrong pattern. The Michigan Court of Appeals decided a case in May 2020, Saleh v. Safeco Insurance Co. of Illinois, in which a driver reported being the victim of a hit-and-run. The insurer's expert inspected the Jeep and downloaded its EDR. According to the State Bar of Michigan's summary of the opinion, the data showed the vehicle was twice driven forward into an object from a stopped position at low but increasing speeds, with the last collision about a minute after the one before, and neither collision had deployed the airbags. The expert concluded the collisions were intentionally staged. The court upheld summary judgment for the insurer under the policy's fraud provision, noting that the policyholders did not dispute the EDR's accuracy and did not submit an expert affidavit of their own (State Bar of Michigan e-Journal summary, No. 73178).

Two lessons come out of that case. The recorded data was persuasive because it conflicted with a specific, checkable account. And the policyholders lost at an early stage in part because they offered nothing to contradict it.

Does the data match the injury or the damage claim?

A belt that was not buckled one second before the event may matter to an injury claim in a state that recognizes a seat belt defense (the seat belt defense: when not buckling up cuts your claim). A recorded change in speed far smaller than the damage estimate implies can prompt a second look at the repair bill. Neither is proof of anything; both are reasons to ask.

Is a stolen or burned car's history consistent?

The National Insurance Crime Bureau's vehicle theft investigation guide, a 2018 publication written with insurer contributors for claim professionals and law enforcement, lists the recorder among the items to check when a stolen car is recovered. It notes that not every vehicle has one, that reading it needs special equipment that is not readily available, and that the owner's permission may be required (NICB Vehicle Theft Claim Investigation Guide). The companion article how insurers investigate a stolen car claim walks through the rest of that process.

A worked example with invented numbers

The figures below are made up to show the method; they are not a real case and not typical results.

Say a driver reports that her sedan was struck while parked on a street overnight and that the other driver left. The estimate is $6,800, and the car has a $500 deductible. The adjuster notices damage on the front of the car, not the side that faced the street, and asks for a download with her written consent. The report shows one non-deployment event with these entries:

Those entries do not prove fraud on their own. A family member might have borrowed the car and hit a post without saying so. What they do is replace a vague dispute with a specific question: who was driving the car at 9 mph with the throttle open, and why does the report say it was parked? The adjuster refers the file to the special investigations unit, who will probably schedule a recorded statement or examination under oath. See My claim is under SIU review: what to expect and your rights for how that process runs and what your rights are in it.

If the driver has an explanation, such as having rolled the car out of a tight spot the previous evening and bumped a bollard, she is better off giving it early and exactly, with a photo of the bollard, than letting the report speak alone. The line between a misremembered detail and a false claim is discussed in claim padding vs fraud: where the legal line falls.

Reliability: what a lawyer or expert will question

An EDR report is persuasive because it is a machine record, but every part of the chain can be challenged. Points that come up:

1. Indicated speed is not ground speed. The regulation calls the field vehicle-indicated speed. Wheel-speed sensors can read high or low under hard braking, wheel lock, skidding or non-standard tire sizes. Experts treat the number as an estimate with a margin.
2. Change in speed is not impact speed. The crash change in speed depends on both vehicles' masses and the angle of the hit. The rule allows a 10 percent error on it.
3. A short window shows little about behavior. With 5 seconds of pre-crash data on most cars, a recorder cannot show that a driver was distracted for a minute or drifting for a mile.
4. Overwriting. For airbag-deployment crashes the module must lock the record. For lesser events, the rule allows the manufacturer to overwrite an earlier non-deployment event with a newer one if no free memory remains (49 CFR 563.9). So the absence of the event you care about may mean a later bump erased it.
5. Interpretation. A report is raw channels, and an engineer decides what they mean. Two experts can reasonably disagree, and the rule's complete-file-recorded field tells the reader whether the unit itself believes the event was fully stored.

An honest claimant should treat the device like any other evidence: ask for the report, have it reviewed if the stakes are high, and keep the module intact until you have decided whether you want your own download.

How EDR data differs from telematics and connected-car data

The EDR is a crash snapshot held in the car and protected by a federal ownership rule. Insurer telematics data is a continuous trip record held by the insurer or its vendor under your program agreement, and connected-car and infotainment data sits with the manufacturer, a service provider or the car itself, under contracts and state privacy law. The Claims Journal authors say infotainment and telematics units can store GPS tracklogs for the preceding days, and sometimes weeks or months, which lets an investigator test the date, time and place of a reported loss; a car that left a repair shop hours before it was supposedly hit while parked is the kind of mismatch they describe. Insurer program data has its own rules, covered in can your telematics data be used in a claim or a lawsuit?. Footage that supports your account is often the most persuasive evidence of all; see dashcam footage and scene evidence: what to capture and keep.

If your insurer asks for your EDR data

1. Ask what exactly is being requested and under what authority. Is it a request for consent, a demand under the cooperation clause, or a court order? The answers lead to different next steps.
2. Check your state. Your state's EDR statute and its insurance department can tell you whether insurers may require consent. Oregon and North Dakota, described above, are examples of states that say they may not, in different ways.
3. Put the scope in writing. If you consent, limit it to this claim, this vehicle and this event, and ask who will receive the report.
4. Ask for a copy of the report and the name of the technician and tool.
5. Keep the module as it is. Do not authorize a repair that replaces the airbag control module until you know whether you want an independent download. If the car is a total loss, ask the adjuster to hold it before it goes to salvage (total loss paperwork: titles, powers of attorney and release forms).
6. Consider your own expert if the claim is large or an accusation has been made. A report favorable to you is only useful if someone can explain it.
7. Get legal advice before declining or when fraud is alleged. A licensed attorney or the state insurance department can tell you where the line falls in your state.

Common questions

Can an insurer deny my claim just because the EDR contradicts me?

A contradiction between the recorded data and your account can lead to an investigation and, if a statement was material and false, to a denial under the policy's fraud provision, as in the Michigan case above. It is a decision based on the whole file, not a single field, and the NAIC model claims rule calls for a denial to be in writing and to cite the policy provision relied on; check your state's version. If you think the reading is wrong, challenge it with your own evidence and ask the state insurance department to review the denial.

Does an EDR show who was driving?

No. It does not identify the driver. It does record whether the driver's belt was buckled and, if the vehicle logs it, steering and pedal inputs, but nothing in it names a person. Who was at the wheel is established by witnesses, phone and camera evidence, and statements.

Can I get my own data to prove I was not at fault?

Usually yes, because it is your data. A qualified technician can image the module, and the report can support a fault argument; it has the most weight when it matches independent evidence such as photos, skid marks and witness accounts. See how fault is determined in a car accident for how fault is decided overall.

Key takeaways

• An event data recorder holds seconds of speed, braking, throttle and belt data around a triggered impact, and the federal baseline for cars built before September 2031 is only the 5 seconds before it.
• The owner or lessee owns the data. An insurer generally needs consent or a court order, and states such as Oregon and North Dakota bar insurers from making consent a condition of coverage or payment.
• Investigators use the data to test whether a claim fits the physical crash, as in the 2020 Michigan case where it showed two staged collisions, and an empty record is a reason to keep looking, not proof of fraud.
• The data has real limits: indicated speed, a 10 percent tolerance on change in speed, a short window, and overwriting of minor events.
• If your insurer asks for a download, find out the legal basis, limit the scope in writing, keep the module intact, and get advice from your state insurance department or a licensed attorney before you refuse.