Every Crash Test Dummy in the World Is Sitting in a Position You Won’t Be In

A Hybrid III crash test dummy costs about $150,000 and has one job: sit in a precisely defined upright position (spine vertical, pelvis centered, hands in lap), absorb a crash pulse, and report what broke. Every Top Safety Pick, every five-star NCAP score, every Good or Poor sticker you’ve ever trusted was measured with the dummy in that position. In a modern AEB-equipped vehicle, you will not be in that position when the crash arrives.

Automatic emergency braking decelerates at up to 1g. It fires late, typically under a second before impact, and during that fraction of a second your head pitches forward by 100 to 400 millimeters, depending on your muscle tone, belt slack, and whether you were braced or mid-sip of coffee.^[1] Oxford’s Transportation Safety and Environment journal measured 188mm of forward head displacement: the occupant’s chest was already making premature contact with the airbag before the crash pulse even began, increasing thoracic compression.^[2] The bag deployed into a ribcage that was closer and moving faster than anything the restraint system was calibrated for.

Three SAE papers published in 2026 dissect this gap. Paper 2026-26-0019 puts it flatly: regulators worldwide evaluate the pre-crash phase and the crash phase independently, with no integration, and the Hybrid III dummy lacks the bio-fidelity to simulate spinal articulation or muscle contraction during emergency braking.^[3] The dummy just sits there. You won’t, because 1g of deceleration and Newton’s first law will have rearranged your posture before the bumper touches anything.

Chalmers University’s SAFER lab quantified the tradeoff with a human body model that has active musculature. At 80 km/h without AEB: 100% probability of two or more rib fractures, 71% concussion risk. With half a second of AEB braking reducing speed to 69 km/h, concussion risk fell to 31%, a massive net positive that nobody in the field disputes.^[1] But when the researchers added seatbelt pre-pretensioning to pull occupants back toward the nominal position, concussion risk crept up again because the higher belt forces delayed airbag contact, producing higher bag pressures at the moment the head arrived. Fix the position problem, introduce a pressure problem.

A Nature study confirmed that every AEB strategy they tested produced “a dislocated state of the driver,” though active seatbelts reduced forward displacement significantly.^[4] They also found that delaying airbag ignition by 20 milliseconds, giving the AEB-lurched occupant a moment to settle, cut injury severity by 39%. Twenty milliseconds. That is the scale of timing the current testing protocol does not account for. NHTSA’s AEB mandate (FMVSS 127, September 2029) specifies when AEB must activate and how much speed it must shed; it says nothing about testing crashworthiness with the occupant in the position AEB put them in.^[5]

Counterpoint, at full strength: SAE paper 2026-01-0556 tested AEB events with actual Hybrid III dummies and human volunteers at 15–50 mph and found that belted occupants were not displaced enough to be out of position.^[6] But that study tested AEB without a subsequent crash. The question is not whether braking alone moves you. It is whether braking followed immediately by a barrier impact produces the same occupant kinematics as impact from a standing start, and no test protocol anywhere in the world asks that question.

By 2029, AEB will be mandatory on every new vehicle sold in the United States, and 77% of 2026 models already meet the updated IIHS standard.^[7] Millions of crashes will never happen, and millions more will happen at lower speeds, which is genuinely good news. But the crashes AEB reduces without preventing — system fires, sheds 11 km/h, car hits anyway — are exactly the crashes where the gap between the dummy’s tested position and yours will matter most.

Source: Research synthesis from peer-reviewed studies (Frontiers, Nature, Oxford Academic) and SAE technical papers. FARS 2014–2023 data provides fleet context but does not track AEB status. See methodology for caveats.