Robot Causes Trouble, Insurance Pays the Bill?

China News Service, March 17 (Li Ziman) — If a robot crashes or damages something, how is compensation handled? Since the beginning of 2026, property insurance companies have continued to expand into the embodied intelligence insurance sector.

Industry experts believe that as the commercialization of embodied intelligence accelerates, the market for embodied intelligence insurance is expected to become an important emerging growth area for the property insurance industry over the next three years.

Three Major Insurers’ Differentiated Strategies

Currently, embodied intelligent robots are used across various fields such as industrial manufacturing, logistics, warehousing, and commercial services. These devices are highly precise, have high maintenance costs, and frequently face new risks, making traditional property insurance products difficult to meet their specific protection needs.

The China Academy of Information and Communications Technology released the “Development Research Report on Humanoid Robot Industry (2024)” in January 2025, predicting that by 2035 to 2040, humanoid robots will reach Lv3 level overall, with the total market size expected to reach 100 billion to 300 billion yuan.

According to the latest data released in March this year by market research firm QYResearch (Hengzhou Bozhi), global embodied intelligence insurance sales are projected to reach $150 million in 2025, growing to $711 million by 2032, with a compound annual growth rate (CAGR) of 24.9% (2026-2032).

Recognizing this potential, PICC Property & Casualty, PICC Property & Casualty, and Ping An Property & Casualty have been launching customized insurance products since 2025 to cover commercial scenarios involving embodied intelligence, seeking breakthroughs in coverage scope and insurance methods.

A relevant person from PICC Property & Casualty told China News Service that in September 2025, the Ningbo branch launched the country’s first humanoid robot-specific insurance, “Smart Insurance.” This product breaks traditional annual policy limits, supporting daily, weekly, and monthly coverage, covering risks across the entire chain of production, sales, leasing, and use. The person revealed that “Smart Insurance” mainly targets complete machine manufacturers. The company has begun nationwide promotion and has already issued policies in Suzhou.

PICC Property & Casualty mainly offers a dual protection system of “body damage insurance + third-party liability insurance.” A relevant person explained that since 2025, the company has launched specialized insurance solutions covering the entire process from laboratory to commercial use of humanoid robots. Coverage includes damage caused by natural disasters, human error, electrical and network issues, as well as third-party liability caused by accidents.

For example, PICC Property & Casualty insures over 470 humanoid robots leased in the Yangtze River Delta region. If damaged due to collision or transmission failure, each robot can receive up to nearly 400,000 yuan in repair compensation.

Ping An Property & Casualty also extends its coverage to leasing. In early January 2026, Ping An Property & Casualty, in cooperation with Shanghai Electric Financial Leasing Co., Ltd., launched an “Insurance + Leasing” policy for embodied intelligent robots, spreading the premium cost into the rent to reduce upfront investment pressure for companies. This product breaks through traditional hardware coverage limitations, including third-party liability, product quality liability, and information leakage liability, utilizing AI algorithms for dynamic risk pricing and real-time risk control.

Challenges of High Risks and Data Scarcity

As a new category of technology insurance, the industry still faces many development bottlenecks.

A relevant person from PICC Property & Casualty told China News Service that compared to traditional corporate property insurance, risks associated with humanoid robots have new features: first, risk sources are more complex, including natural disasters, hardware failures, algorithm bias, and system vulnerabilities; second, responsibility attribution is more complicated, involving hardware manufacturers, software developers, or users; third, emerging risks such as cybersecurity, algorithm security, and information leaks are not covered by traditional corporate insurance.

Wang Guojun, a professor at the Insurance School of the University of International Business and Economics, believes that humanoid robots face systemic risks like algorithm decision errors and cyber hijacking, with no historical claims data, multiple technological pathways, and frequent software updates. The core contradiction is the mismatch between “high risk, rapid iteration, lack of data” and “insurance needs large numbers, stability, and quantifiability.”

Bai Wenxi, chairman of Zhonghe Kunlun Asset Management Co., Ltd., pointed out that as embodied intelligence scenarios continue to penetrate, professional insurance protection is expected to become an important growth point in the property insurance field. However, the industry currently faces three core challenges: first, risk identification and data scarcity render traditional actuarial models ineffective; second, responsibility boundaries are blurred, involving multiple parties such as algorithm developers, hardware manufacturers, and operators; third, the combination of long-tail and systemic risks, with robot clusters potentially experiencing “cascading failures,” breaking the foundation of the law of large numbers.

Regarding pricing, a relevant person from PICC Property & Casualty said that there is almost no historical claims data. Currently, their approach is “dynamic assessment” combined with “cross-industry cooperation.” They reference data from mature categories like industrial and service robots, considering new robot safety levels, operational scenarios, and maintenance costs to improve accuracy dynamically. They also collaborate deeply with robot manufacturers and industry chain companies to conduct “health checks” on robots, breaking down risks into “brain,” “small brain,” and “body,” to establish a dynamic risk assessment system.

Wang Guojun believes that insurers should move forward by building shared data platforms, implementing dynamic rates, modular clauses, and hybrid pricing models combining simulation and real-machine testing, transforming “uncertainty” into tradable “risk products.”

To address industry challenges in embodied intelligence insurance, Bai Wenxi suggests creating a “data-ecosystem-regulation” three-in-one plan: establishing a shared data alliance, promoting “insurance + services + technology” in the ecosystem, and setting industry standards and mandatory insurance regulations. Short-term, pricing should be scenario-based and tiered; long-term, adopting Usage-Based Insurance (UBI) models focused on risk reduction rather than loss compensation, turning insurance from a cost center into a value partner for shared risk governance.