China Industrial Cooperation Association
Shanghai Federation of Industrial Economics
Shanghai Federation of Economic Organization
Industrial and Information Technology Equipment Engineering Research Institute (Beijing) Co., Ltd
Green Industry Enerey Conservation Branch, CICA
Golden Conference & Exhibition Group
Shanghai Berrick Exhibition Co., Ltd
A humanoid robot tips over in a shopping mall and pins a bystander's foot. A hospital assistant misreads a dosage instruction and bumps a patient's IV stand. A factory unit, its vision model briefly blinded by a glare, drives a forklift into a wall. None of these are science fiction anymore. Shanghai International Humanoid Robot and Robotics Industry Chain Exhibition 2026 (HRIE 2026), to be held December 9–11, 2026, at the National Exhibition and Convention Center (SNIEC), Shanghai, is set to become the stage where manufacturers, operators, and insurers confront a question the industry can no longer dodge: when a humanoid robot causes harm, who pays? As machines move out of the lab and onto the shop floor, into hospitals, and alongside pedestrians with no safety fence in between, the third-party liability that once lived only in spec sheets is becoming a real-world balance-sheet item.
For years, humanoid robots were confined to trade-show stages, fenced-off demo zones, and carefully scripted videos. That era is closing fast. According to public reports, humanoids are now being deployed in science museums, commercial malls, factories, medical settings, and inspection rounds — and close-proximity, even fence-free, human–robot collaboration is becoming routine. When a machine weighing 40–80 kilograms stands shoulder to shoulder with a child in a mall or walks an unguarded corridor in a hospital, the margin for error is measured in centimeters, not meters.
The discussion around third-party bodily injury and property damage from operational accidents and safety-distance deviations has moved from academic to urgent. Industry analysts note that the conversation intensified markedly after the 2026 World Robot Conference, as deployments broadened and the gap between demo polish and real-world messiness became harder to ignore. The core problem is not that robots are malicious; it is that they are autonomous, heavy, sensor-dependent, and networked — a combination that creates novel failure modes no one has insured at scale before.
Here is the uncomfortable logic that underwriters keep repeating: insurance cannot price what the law does not define. To sell a policy, an insurer must answer four questions — what is the peril, who is the insured, who is liable, and how much will it cost. If a judge, regulator, or jury cannot reliably assign fault among the manufacturer, the software developer, the data supplier, the on-site operator, and the end user, then no one can compute a premium. Legal certainty is the precondition for insurability, and insurability is the precondition for scaled deployment.
This is not a theoretical bottleneck. Humanoid robots sit at the intersection of product liability, professional liability, cyber risk, and pure operational accident. In most jurisdictions, a traditional public-liability policy was written for spilled coffee and slippery floors — not for an algorithm's split-second decision or a remote firmware update that changes how a machine behaves. Without new legal frameworks and new policy wordings, the fastest, most capable robots may simply be uninsurable outside narrow Pilots, which in turn slows the very commercialization everyone is racing toward.
China is widely regarded by industry analysts as one of the fastest-moving markets for humanoid deployment, and its regulators have moved early to close the insurance gap. In March 2026, China's National Financial Regulatory Administration (NFRA) and three other ministries jointly issued guidance that explicitly named artificial intelligence and robotics as priorities for technology-insurance innovation, encouraging the use of insurance mechanisms to share emerging risks.
That policy signal unlocked a wave of product design. The logic was straightforward: if the state says these risks are legitimate to underwrite, insurers can stop waiting for perfect data and start writing coverage, collecting the loss experience that will eventually let them price it properly.
China's leading property-and-casualty insurers have each staked out a distinct wedge of this market.
PICC (People's Insurance Company of China). According to public reports, PICC has built a two-track system — "physical-damage cover for the robot本体" plus "third-party liability cover" — and has already provided protection for a first batch of roughly 470 humanoid robots on a Yangtze River Delta leasing platform. The dual structure matters: it treats the machine as both an asset to be protected and a source of harm to be insured against.
CPIC (China Pacific Property Insurance). CPIC launched what public reports describe as the country's first dedicated humanoid-robot insurance product, branded "Ji Zhi Bao / 机智保." Its notable innovation is breaking the traditional annual-policy constraint: it supports coverage priced by the day, week, or month. For a robot that may be leased for a single weekend event or a two-week warehouse pilot, that flexibility is the difference between a sellable product and a non-starter.
Ping An Property & Casualty. As reported, Ping An's relevant business spans medical, energy, consumer, and infrastructure tracks, with cumulative insured sums reportedly exceeding RMB 170 million. The breadth signals that insurers see humanoid risk not as a niche but as a cross-sector line of business.
The timeline kept accelerating. At the China International Fair for Trade in Services (CIFTIS) on September 11, 2026, CPIC debuted an insurance product for key components of embodied-intelligence robots, focused on the dexterous-hand direction, with a first policy reportedly landing the same day. Coverage is migrating from the whole machine to the most failure-prone, hardest-to-replace parts — a sign of a market maturing past generic promises toward engineered, component-level risk transfer.
Frameworks only matter when they pay out. A concrete case gives the abstract debate a grounding point. According to public reports, a robot on the Qitianzu (擎天租) leasing platform accidentally toppled during use and was damaged; the owner eventually received a PICC claim payout of approximately RMB 5,976.87. Publicly described as one of the earlier practical payouts in China's robot-leasing insurance space, the case marks insurance stepping off the page of policy clauses and into the full commercial operating loop — underwriting, incident, adjustment, and settlement.
Note the modest size of the number. The significance is not the amount; it is the precedent that a real loss was adjudicated and paid under a real policy. That single transaction generates the loss data that a thousand demos never will.
The temptation is to treat robot insurance as a familiar line with a new label. It is not, and the reasons are structural.
Composite causation. A single harmful event can blend hardware breakage, algorithm bias, system downtime, cyber attack, and data leakage. Traditional policies carve these into separate silos — property, professional liability, cyber — that may each deny coverage by pointing at the other. A robot that falls because a hacker spoofed its sensor is simultaneously a cyber event, a product failure, and a public-liability accident.
Hard-to-define liable party. Responsibility sprawls across equipment manufacturing, software development, data provision, on-site operation, and the end user. In a fence-free factory, is the harm attributable to the robot maker's joint tolerances, the model developer's training data, the facility's safety protocol, or the operator who chose to disable a warning? Until law and contract allocate that chain, insurance sits on uncertain ground.
Undefined perils. As noted, classic public-liability wording does not specifically contemplate robot operation, algorithm behavior, or human–robot interaction. Insurers are therefore drafting bespoke endorsements, but each bespoke wording is a small legal experiment whose interpretation is untested in court.
These are precisely the frictions that, absent resolution, keep premiums high and capacity thin.
For an English-reading audience, the global picture is less a coordinated system than a patchwork — and the contrast sharpens the central thesis.
United Kingdom. As reported by CETAS in a March 2026 report, the UK still has no dedicated robot-liability law, even as deployment grows faster than the regulatory response. The gap between market speed and legal scaffolding is exactly the insurability trap described above.
European Union. The EU's AI Act entered into force in August 2026, and its Machinery Regulation is scheduled to apply from January 2027. Brussels is building the legal certainty — classification, obligations, conformity — that insurers say they need before they can confidently price robot risk across the bloc.
United States. Washington's relevant 2026 action has leaned toward trade-security framing, notably the FCC Covered List updated in July 2026, rather than a comprehensive robot-liability statute. The result is a market where coverage is emerging product-by-product, state-by-state, and lawsuit-by-lawsuit.
Industry expectations, as relayed by analysts, are that syndicates at Lloyd's of London or specialist technology insurers will bring dedicated humanoid-robot products to market within an 18–24 month window. The signal from the grassroots is more telling still: a question on Hacker News asking whether one can insure a party humanoid robot surfaced the raw exposure in leasing and event scenarios — a coverage blank and an actuarial data void hiding in plain sight.
Step back and the pattern is consistent across every market: regulation and insurance are not sequential, they are symbiotic. Law creates the definable liability that lets insurers write; insurance creates the distributed financial cushion that lets regulators permit deployment without taxpayers or victims bearing unbounded risk. Where one lags, the other stalls. China's March 2026 directive and its insurers' rapid product launches are one half of the loop; the EU's statutory timetable is the other. The UK and US, by contrast, illustrate what partial certainty produces — innovation outrunning the safety net.
For operators and lessors, the practical takeaway is to treat insurance as a design input, not an afterthought. Choose platforms whose manufacturers accept contractual responsibility, demand component-level coverage where parts are the weak link, and favor flexible terms (daily/weekly/monthly) that match how robots are actually deployed. For policymakers, the lesson is that vague intentions do not create insurability — specific allocation of fault does.
The humanoid is no longer a prototype with a press release; it is a coworker, a greeter, a nurse's aide, and a warehouse runner. Every one of those roles carries the realistic chance of a third party being hurt or a piece of property being damaged, by accident, by bias, or by attack. The insurance and liability framework is the safety net that turns "demonstration" into "deployment at scale" — and until legal certainty lets insurers price the risk, the most advanced machines may be the ones least able to leave the building.
That is why the conversation cannot stay in white papers. At HRIE 2026, held December 9–11, 2026, at the National Exhibition and Convention Center (SNIEC), Shanghai, manufacturers, operators, and insurers will share the same floor to negotiate exactly this: how to underwrite the reality of humans and robots coexisting without a fence between them. The exhibition is where the industry's most pressing question — who pays when the robot causes harm — will move from debate toward a working answer.