1979. $15 million. 2025.

Three numbers. One story. Forty-six years of distance.

1979. On January 25, 1979, Robert Nicholas Williams, a 25-year-old factory worker at Ford Motor Company's Flat Rock Casting Center in Michigan, was struck and killed by the arm of a one-ton robotic transfer vehicle. He was the first recorded human killed by a robot. The system, built by the Unit Handling Systems division of Litton Industries, was designed to retrieve castings from high-density storage shelves. When the robot produced erroneous inventory readings, Williams was asked to climb into the racks to retrieve parts manually. The robot's arm struck him from behind. He was killed instantly.

Williams lay dead for thirty minutes before coworkers found him. The robot continued operating. No alarm sounded. No sensor detected a human in the workspace. No safety system intervened. In 1979, none of those capabilities existed.

$15 million. In August 1983, a Wayne County Circuit Court jury found Litton Industries liable for negligence in designing, manufacturing, and supplying the storage system — and for failing to warn operators of foreseeable dangers. The jury awarded Williams's estate $10 million, the largest personal injury verdict in Michigan history at that time. In January 1984, the award was raised to $15 million. Litton ultimately settled for an undisclosed amount without admitting negligence.

Paul Rosen, the family's attorney, told the court: "We have to be very careful that we don't go backwards to the kind of notions we had during the Industrial Revolution that people are expendable."

Williams was far from the last. A 2023 NIOSH study by Larry Layne, published in the American Journal of Industrial Medicine, analyzed 26 years of Bureau of Labor Statistics data and identified 41 robot-related fatalities in the US between 1992 and 2017. Of those, 85% of victims were male, 46% occurred in the Midwest — concentrated in Michigan and Ohio's auto manufacturing corridor — and 78% involved a robot striking the worker while operating under its own power. The most common context: maintenance. Workers performing unjamming, sensor cleaning, or troubleshooting accounted for the majority of fatal incidents.

2025. Forty-six years after Robert Williams, the first insurance products built specifically for robots finally appeared — and not in the United States or the European Union. In October 2025, China Pacific Insurance launched "机智保", the first policy written exclusively for humanoid robots, covering the machine itself and third-party liability, sold by the day, week or month. Ping An and PICC followed within weeks. In the specialty market, Axis Insurance's autonomous robotics programme began underwriting what carriers had refused to touch: bodily injury from an AI perception failure, a cyberattack that seizes a robot's controls, and production losses when a software update takes a fleet offline. Relm's PONTAAI, launched in January 2025, covers bodily injury and property damage across the wider AI stack. The movement carried into 2026 — Zurich and YAS launched embedded robotics cover in Hong Kong in June.

That third clause is the one that matters. It is the first time an insurer has priced the failure mode that defines fleet robotics — one bad update, every robot at once.

But these are first movers writing policy ahead of the data, not a mature market. There is still no public failure-rate baseline, no standardised diagnostic-coverage reporting, and no decades of actuarial history to price against. The first robot killed a man because it could not sense him. Today's robots sense everything — and a single flawed update can disable every safety function across a thousand of them simultaneously. The failure mode changed in 1979. The insurance is only now, forty-six years later, beginning to catch up.