Physical AI poses deeper threat to insurance than generative AI, EY warns
Autonomous vehicles and humanoid robots will force insurers to overhaul their core business models in ways that generative AI has not, according to EY's insurance consulting leader for the Americas.
The shift stems from a fundamental change in how liability flows through the economy. Traditional insurance has anchored premiums to human operators. Physical AI systems distribute risk across manufacturers, software vendors, and vehicle owners-a complexity that existing underwriting practices cannot easily absorb.
Chris Raimondo said insurers are only beginning to grapple with these implications. "Physical AI could potentially be more disruptive to the insurance industry as a whole than digital AI," he said. Most insurer activity today focuses on internal efficiency gains from generative AI in underwriting, customer service, and fraud detection.
The liability shift: From drivers to systems
Consider autonomous vehicles. Personal and commercial auto insurance have long priced risk based on driver behavior. As self-driving technology scales, liability may extend across the vehicle manufacturer, the AI software platform, and the fleet operator or owner.
The US National Highway Traffic Safety Administration data shows advanced driver-assistance and autonomous systems already deployed across millions of vehicles, with manufacturers reporting billions of autonomous miles in testing and limited commercial operations.
Underwriting will need to account for "software versioning, sensor capabilities, and data generated by the vehicle," Raimondo said. Premiums will shift from static annual rates to dynamic pricing adjusted continuously based on real-time system data.
New products and claims models emerging
Insurtech firm Lemonade already launched a usage-based policy that adjusts pricing based on whether a vehicle operates in autonomous or human-driven mode. This signals broader movement toward continuous premium recalibration.
Opportunities extend beyond vehicles themselves. Insurers are exploring coverage for the infrastructure supporting physical AI-charging networks, data centers, smart roads, and edge computing facilities.
Claims processing faces equally significant change. Today's models rely on human investigation and adjudication. Physical AI will generate vast quantities of machine data that claims teams must interpret.
"Claims will rely more heavily on sensor data, OEM logs, and software diagnostics," Raimondo said. "Evaluating software faults will become a core component of claims adjudication."
The timeline remains uncertain
Global AI spending is projected to exceed $500 billion by 2027, with growing investment in physical applications. Goldman Sachs estimates the humanoid robotics market alone could reach $150 billion by 2035, driven by labor shortages and machine learning advances.
Adoption rates for autonomous vehicles and robotics vary significantly across regions. As investment accelerates and regulatory frameworks develop, insurers face mounting pressure to adapt their entire value chains-not just their products.
Raimondo said the industry remains in early stages. "Adoption and investment are increasing, and insurers are beginning to recognize that their business models, not just their products, may need to evolve significantly."
For insurance professionals, understanding AI for Insurance and AI Agents & Automation will become essential as these systems move from testing into widespread deployment.
Your membership also unlocks:
