Koop says robot trust is rising as insurance costs fall

Koop published a new report built on five years of insurance data from 392 U.S. robotics companies, arguing that Physical AI is becoming easier to insure and cheaper to trust. The 35-page study says capacity has surged, losses are concentrated in system-level failures, and safety controls are lowering prices even as many companies fail to prove they have them.

Why it matters: - Robotics adoption depends on trust from customers, landlords, cities and defense buyers who often require liability coverage before a robot can operate. - Koop’s report says insurance pricing now reflects how much the market believes a dollar of robot revenue will cost in future claims. - The findings suggest robot liability is becoming more capacity-rich and less expensive as underwriters gain more loss history. - The report also points to a coming shift in standards, compliance and policy wording that could shape how physical AI is deployed.

What happened: - Koop published IN BOTS WE TRUST 2026, a 35-page report on robot risk built from insurance quotes, policies, claim files and underwriting applications. - The report covers 392 U.S. robotics companies, including developers, operators and software companies that enable robotics. - The analysis spans policy years 2021 through 2025. - Koop says that in each of the last two years, between one in five and one in four U.S.-headquartered robotics companies came to its program for a quote or policy. - The report is free at the 2026 report.

The details: - Robotics losses were described as a top-1% insurance risk, with the program’s loss ratio outperforming the best-in-class result among the 25 largest U.S. property and casualty groups. - Commercial auto liability, the closest analogue for moving robots, posted a 71% loss ratio in the first half of 2025. - Aggregate liability limits deployed for robotics rose from $80 million in 2022 to $572 million in 2025, including $452 million per occurrence. - Premium per dollar of revenue in 2025 fell to 31% of the 2022 level, a 69% drop in three years. - The median policy rate declined 40%. - The report describes a price ladder with the autonomous mobile robot as the benchmark at 100, robotics software at 6, drones at 50, robot arms at 172, delivery robots at 191, and machines that clean, dig and move things in occupied spaces at 360 to 700. - Cyber incidents account for 34% of loss dollars. - Incidents in which a member of the public was hurt account for 12% of loss dollars. - Only 4% of companies have ever reported an incident. - The leading root cause is how a robot was deployed and supervised. - 87% of applicants report a safe-stop system, 74% report cybersecurity measures and 65% report a safety framework or third-party evaluation. - Companies reporting a control pay 25% to 65% less per dollar of revenue. - Underwriters credit a named standard or safe-stop in only 7% of reviews. - A small minority hold a cybersecurity compliance framework, and 6% publish a safety page. - The report maps standards that are expected to shape pricing, including ISO 10218:2025 and its U.S. adoption as ANSI/A3 R15.06-2025, R15.08 for mobile robots, the revised ISO 13482 for service robots, UL 4600 for autonomous products, the first standard in development for legged and humanoid robots, and the EU Machinery Regulation that applies from January 2027.

Between the lines: - Koop is framing robot adoption as a trust problem, not just a hardware problem. - The report suggests the strongest pricing advantage may go to companies that can prove their safety controls, instead of merely claiming them. - That gap matters because underwriters appear to reward evidence far less often than companies report having the controls. - The report’s analysis implies that system-level failures, especially deployment and supervision, remain the main underwriting concern rather than robot hardware alone. - Samuel Reeves of FORT Robotics said robot safety has to extend to the entire system around the machine, with verifiable controls, secure communications and human oversight turning unknown risks into manageable ones. - Grayson Brulte of The Road To Autonomy said robot liability is becoming cheaper, more capacity-rich and underwritten at scale, calling that a missing layer of the Autonomy Economy.

What’s next: - Koop says it plans to publish the report annually. - The report makes five 2027 predictions: rates keep falling before leveling off, demand for excess capacity rises with fleet size, cyber becomes the formal loss of record for robotics, warranties replace exclusions in policy wording, and safety standards plus compliance reports become underwriting inputs. - The company expects future pricing to be shaped by new and updated standards, including the EU Machinery Regulation in January 2027.

The bottom line: - The robotics insurance market is getting cheaper, but the report says the winners will be the companies that can prove their safety and cybersecurity controls with evidence, not marketing.