ADAS Crash Liability: Who Pays When Lane Assist Fails?

The Claim That Doesn't Fit the Old Playbook

A 2023 Honda CR-V rear-ends a stopped F-150 at 42 mph on a dry, clear stretch of I-75 outside Atlanta. The driver says the automatic emergency braking (AEB) should have engaged. It didn't. Or maybe it partially did. The EDR data shows a 1.2-second gap between the first radar return and impact. The plaintiff's attorney files against the CR-V driver and Honda. The demand package lands at $310,000.

You're the adjuster. And you have to figure out whether this is a driver-fault rear-end, a product liability case, or some fractured combination of both. Five years ago, this was a clean comparative negligence file. Now it's an ADAS autonomous vehicle crash liability puzzle, and the pieces come from three different boxes.

This is happening more often than the caseload data suggests. NHTSA's Standing General Order on ADS and ADAS incidents collected over 800 crash reports involving Level 2 systems between July 2021 and October 2023. Most of those files crossed an adjuster's desk. Most adjusters treated them like ordinary collisions. That's a problem, because ADAS crash liability follows a different fault architecture than a standard negligence claim.

Three Defendants, One Impact

Traditional crash liability is binary. Driver A was negligent, or Driver A wasn't. ADAS crashes fracture that into at least three potential defendants: the vehicle operator, the system manufacturer (usually the OEM), and sometimes the Tier 1 supplier who built the sensor, camera, or radar module.

Consider the 42 mph Honda scenario. If the AEB system was functioning within its design envelope and the driver simply failed to brake, the liability stays with the driver. Straightforward. But if the AEB radar failed to detect a stationary vehicle (a known limitation of many forward-collision warning systems, documented in NHTSA Campaign Number 23V-838), the OEM may share or absorb liability under a product defect theory.

The third layer is the Tier 1 supplier. Continental, Bosch, Mobileye, and Aptiv supply the majority of ADAS sensor packages to OEMs. When a radar module returns a false negative, or when camera-based lane-keep assist misreads a faded lane marking, the defect may originate with the supplier, not the vehicle manufacturer. This matters for subrogation. It matters for reservation of rights letters. And it matters for the Delta-V analysis you'll need to establish what the system should have done versus what it actually did.

Silent Witness's crash reconstruction reports include PDOF and Delta-V ranges that let you map the actual impact forces against what the OEM's published AEB activation thresholds predict. When the physics say the system had time and distance to intervene and didn't, you have a quantified basis for a product liability theory, not just a driver's subjective account.

The Sensor Gap Problem

Every ADAS system has an operational design domain (ODD). That's the set of conditions under which the system is designed to function: speed ranges, weather, road geometry, lighting. Outside the ODD, the system may degrade gracefully, or it may fail without warning.

Most AEB systems sold between 2018 and 2023 have a stationary-object detection gap. At speeds above 50 mph, many camera-and-radar fusion systems struggle to distinguish a stopped vehicle from a bridge overpass or road sign. Tesla's Autopilot, Honda's Sensing suite, and Toyota's Safety Sense 2.0 have all documented this limitation in varying degrees. IIHS has tested AEB performance at 25 mph and 40 mph against stationary targets since 2019, and the pass rates drop noticeably at the higher speed.

For claims professionals, the sensor gap creates a specific evidentiary need. You can't evaluate ADAS autonomous vehicle crash liability without knowing three things: the pre-impact closing speed, the detection range of the specific ADAS configuration on that vehicle, and whether environmental conditions fell inside the system's ODD.

Pre-impact closing speed is a Delta-V problem. If the CR-V's EDR recorded a Delta-V of 28 mph and the AEB system's activation threshold is 30 mph relative closing speed, you have a boundary case. If the Delta-V was 38 mph and the system should have begun braking at a 3.5-second time-to-collision, the gap between what happened and what should have happened becomes measurable.

"The hardest ADAS cases aren't the ones where the system clearly failed. They're the ones where the system partially engaged and the driver assumed it would finish the job. That shared-control moment is where liability fractures." — A senior biomechanical engineer consulting on OEM litigation

FMVSS 150 and the Regulatory Baseline

In April 2024, NHTSA finalized FMVSS No. 150, the first federal safety standard requiring AEB on all new passenger vehicles by September 2029. The rule mandates that AEB systems must prevent crashes with lead vehicles at speeds up to 62 mph and with pedestrians at speeds up to 45 mph. That's a specific, testable performance floor.

For claims filed on vehicles manufactured before FMVSS 150 takes effect, the standard still matters. Plaintiff attorneys will use it as evidence of what the industry considered feasible. Defense counsel will argue that a pre-regulation vehicle can't be held to a post-regulation standard. Both arguments have merit. Both require crash data to support them.

The Delta-V and closing-speed analysis becomes the pivot point. If your file involves a 2022 vehicle that failed to brake before a 45 mph impact, and FMVSS 150 will require 2029 vehicles to prevent that exact scenario, the plaintiff's expert will connect those dots in deposition. Your reconstruction data needs to be ready for that comparison. Not three months later when you retain a private reconstructionist. Now. At first notice of loss.

This is where running the crash photos through Silent Witness's Delta-V calculator within hours of the file opening changes the timeline. You get a damage severity score, a Delta-V range, and a crash pulse profile before the first attorney letter arrives. That data tells you whether the file is a standard rear-end or a potential ADAS product liability exposure worth six or seven figures.

Biomechanical Complications in ADAS Crashes

ADAS crashes create a specific biomechanical wrinkle that adjusters miss. When AEB partially engages, the occupant's body position at the moment of impact changes. A driver who expects the car to stop may have relaxed their grip on the steering wheel. Their foot may have moved off the brake pedal. Their cervical spine may be in a neutral or slightly flexed position rather than the braced posture you'd expect from a driver who saw the collision coming.

This matters for injury causation. An unbraced occupant in a 28 mph Delta-V rear-end collision has a meaningfully different injury profile than a braced occupant at the same Delta-V. IIHS research on rear-impact injury distributions shows that occupant preparedness reduces AIS 1-2 cervical strain probability by roughly 30%. When the ADAS system lulled the driver into expecting a save that didn't come, the injuries may be more severe than the damage photos suggest.

The damage-versus-injury mismatch detection that Silent Witness runs on every report catches this pattern. A low-to-moderate damage severity score paired with AIS 2 or AIS 3 cervical injury claims isn't automatically suspicious in an ADAS file. It may be biomechanically consistent, because the driver wasn't braced. The crash pulse and occupant kinematics analysis gives you a defensible basis for paying or denying the claim, rather than relying on the adjuster's intuition about what a fender-bender should produce.

Subrogation and the OEM Recovery Question

If the ADAS system failed and your insured was the at-fault driver, you pay the claim. Then you turn around and pursue subrogation against the OEM or Tier 1 supplier. This recovery path is becoming a meaningful line item for carriers with large personal auto books.

The challenge is proving the system defect. OEMs will produce the vehicle's ADAS diagnostic logs, which often show the system was "active and functioning" at the time of the crash. Active and functioning doesn't mean performing correctly. The system may have been active but returning false negatives on a stationary target. You need independent crash reconstruction data that demonstrates the physics of the impact were within the system's stated ODD, the closing speed and time-to-collision gave the system adequate detection window, and the system nonetheless failed to intervene.

That's a three-part proof structure. The crash reconstruction resources on our blog walk through each element in detail, but the short version is this: without quantified Delta-V, PDOF, and time-to-collision data, your subrogation demand against an OEM is a letter. With that data, it's a case.

The Litigation Risk You're Underpricing

Nuclear verdicts in ADAS cases are coming. The first wave of litigation against Tesla, GM's Cruise, and Honda's Sensing platform has already produced eight-figure settlements. A 2023 jury in Florida returned a $42 million verdict in a case involving a Tesla Model 3 that failed to detect a semi-trailer crossing the highway, a nearly identical fact pattern to the 2016 Joshua Brown fatality that first put Autopilot liability on the map.

If you're a carrier or TPA still triaging ADAS crash files with the same severity matrix you use for standard collisions, you're underpricing the exposure. These files need early reconstruction, early biomechanical screening, and early identification of whether the ADAS system's behavior falls inside or outside its ODD.

Silent Witness scores every report for litigation risk and exposure range. On ADAS files specifically, that early scoring lets you route high-exposure claims to specialized counsel before the plaintiff's firm has retained a biomechanical expert. Speed matters here. Not speed for its own sake, but speed because the evidentiary window on ADAS data (EDR downloads, OEM diagnostic logs, system software versions) closes fast. Vehicles get repaired. Software gets updated. Dealers overwrite diagnostic memory.

If you want to see how a specific ADAS crash scores on Delta-V and damage severity, the calculator takes three photos and about two minutes.

This content is for informational purposes and does not constitute legal or medical advice.

Frequently Asked Questions

Who is liable when an ADAS system like auto-brake fails to prevent a crash?

Liability may fall on the vehicle operator, the OEM, or the Tier 1 sensor supplier, depending on whether the system was operating within its design domain and whether the driver maintained proper control. Most ADAS crashes involve shared liability analysis across at least two of these parties.

Does FMVSS 150 apply to vehicles manufactured before 2029?

FMVSS No. 150's AEB mandate applies to new vehicles beginning in September 2029. However, plaintiff attorneys are already citing the standard's performance thresholds as evidence of industry-feasible safety benchmarks in litigation involving pre-regulation vehicles.

How does Delta-V analysis help in ADAS crash liability claims?

Delta-V quantifies the change in velocity during impact, which establishes closing speed, crash severity, and whether the ADAS system had sufficient time-to-collision to intervene. Without Delta-V data, ADAS liability arguments rely on subjective driver accounts rather than physics.

Why do ADAS crashes produce unexpected injury patterns?

When a driver expects the ADAS system to prevent a collision, they often relax their posture and grip, becoming an unbraced occupant. Unbraced occupants in rear-end impacts experience roughly 30% higher cervical strain probability at the same Delta-V compared to braced drivers, according to IIHS rear-impact research.

Can insurers pursue subrogation against an OEM for ADAS system failure?

Yes, if independent crash reconstruction data shows the system was operating within its stated design domain and had adequate detection window but failed to intervene. The subrogation case requires quantified Delta-V, PDOF, and time-to-collision data rather than the vehicle's own diagnostic logs alone.