You build machines that sense, decide, and act in the physical world — drones, robots, autonomous vehicles, and the industrial systems that operate alongside people and property. The moment software leaves the screen and begins to move through the physical world, the nature of the risk changes entirely: a software error is no longer a financial loss — it becomes a bodily-injury claim. That calls for a program engineered around the exposure, and a broker who has placed it before.
Artificial intelligence has spent a decade living on screens. It is now moving into the physical world at scale. Industry counts put millions of industrial robots on factory floors worldwide, with hundreds of thousands of new units installed every year. Collaborative robots now share unfenced floor space with human workers. The humanoid robot market, effectively zero a few years ago, is one of the fastest-growing categories in hardware. Commercial drones have grown into a multibillion-dollar global market, and the Federal Aviation Administration's proposed Part 108 rule is set to replace one-off waivers with a standing framework for flights beyond the operator's visual line of sight. Autonomous vehicles have crossed from pilot to paid service — leading operators now run hundreds of thousands of paid robotaxi trips a week and are expanding into new cities.
Every one of these machines carries an exposure that traditional technology insurance was never designed to hold: the capacity to injure a person or damage property on its own. When something goes wrong, the claim rarely stops at one party. It can pull in the company that built the hardware, the company that wrote the autonomy software, and the company operating the machine — all at once. Insuring physical AI well means understanding where the liability actually lands, and structuring coverage so nothing falls through the seams between policies.
Hull, third-party liability, and payload coverage for commercial drone operators and manufacturers — including Part 107 operations today and the beyond-visual-line-of-sight framework arriving under the Federal Aviation Administration's proposed Part 108 rule. Placed with specialty aviation carriers who understand the operational controls these flights require.
Coverage for ground robots, humanoids, collaborative robots, and factory automation. When a machine shares space with your people or your customers, the exposure is bodily injury and property damage — so we lead with product liability and casualty, and we work directly in the safety standards, ISO 10218 and ISO/TS 15066, that underwriters expect you to meet.
Liability, product, and technology coverage for self-driving cars, trucks, delivery robots, and autonomous ground and marine platforms — structured for the manufacturer, the operator, and the software developer behind the system, in a regulatory environment that still varies state by state.
This is the exposure that separates physical AI from software. A robot, drone, or autonomous vehicle can injure a person directly — and when it does, the hardest question is who is at fault: the machine, the code, or the operator. Claims history is thin and courts are still setting precedent, which makes these losses unpredictable and expensive to defend. In one widely reported incident, an autonomous vehicle dragged a pedestrian after another car struck her; the operator ultimately faced a multimillion-dollar settlement with the victim along with federal safety and criminal penalties. We structure liability programs that anticipate multi-party disputes instead of assuming a single defendant.
If you manufacture the hardware, a defect in design, components, or firmware can trigger claims across every unit you've shipped — plus the cost of a recall. The intersection of a physical product with embedded, self-updating software creates coverage gaps that standard product liability policies were not written for. We place product and recall coverage that accounts for the software layer, not the hardware alone.
These are capital-intensive machines. A fire, natural disaster, theft, or in-field crash can destroy equipment that was purpose-built and slow to replace. Drone operators need hull coverage for the aircraft and its payload; robotics and automation companies need property and equipment coverage sized to specialized, high-value hardware.
Most of these systems are connected, remotely monitored, and updated over the air — which means a cyberattack is also a physical-safety event. An intruder can exfiltrate production data or, worse, take control of a machine that moves. Data breaches routinely cost affected companies millions before accounting for any physical damage, which places cyber and technology errors and omissions at the center of the program rather than at its margins.
The rules are being written in real time. The Federal Aviation Administration's Part 108 proposal will reshape how beyond-visual-line-of-sight drone operations are approved. The National Highway Traffic Safety Administration is still developing federal autonomous-vehicle policy while a growing number of states permit driverless commercial operations under their own terms. And the European Union AI Act brings phased obligations for AI embedded in machinery and vehicles, with penalties attached. We build coverage that anticipates regulatory change rather than assuming today's rulebook holds.
Defects in hardware, components, or firmware across every unit shipped, plus recall costs.
General Liability →Bodily injury and property damage caused by a machine in operation — often contractually required per site.
Technology E&O →Failure of the autonomy software, missed performance obligations, and integration errors.
Cyber →Remote compromise, data breach, and the physical-safety fallout of a hijacked machine.
Commercial Property →Physical damage to high-value, purpose-built equipment and facilities.
Drone Liability →Hull, third-party liability, and payload coverage for aviation systems.
There is no list price for insuring autonomous hardware. Premium is built from your specific risk profile, and a few factors move it the most.
A prototype tested on a closed course prices very differently from a fleet operating in public space or a factory floor shared with workers. Units in the field, miles or flight hours logged, and proximity to people all matter.
Underwriters reward evidence. Documented testing, third-party certification, adherence to recognized standards (ISO 10218 and ISO/TS 15066 for robotics; Part 107 or Part 108 controls for drones), and clear human-oversight procedures can materially lower your cost.
If you build the machine, product liability and recall dominate your program. If you operate it, third-party liability and hull or physical-damage exposure lead. Many companies are both, and the program has to reflect that.
Retentions, limits, and how coverage is layered across markets directly affect price. A higher deductible lowers premium but shifts more of each loss to you.
A clean record helps; prior product, liability, or malfunction claims raise the bar. Because loss data in this space is still thin, how you present your risk carries unusual weight.
A driverless vehicle is involved in a pedestrian injury after a human-driven car initiates the collision. Fault is contested across the vehicle operator, the software developer, and the third-party driver; regulatory reporting obligations attach immediately. Real cases in this category have produced settlements in the millions plus separate federal penalties for how the incident was reported. The right program funds a coordinated defense instead of leaving each party to fight alone.
A cobot makes unexpected contact with a worker on a shared line. The investigation turns on whether the installation met ISO 10218 / ISO/TS 15066 force-and-speed limits and OSHA's general duty obligations. Product liability, general liability, and workers' compensation can all be triggered by a single event.
A beyond-visual-line-of-sight flight ends in a crash that destroys the aircraft, its sensor payload, and damages property below. Hull, payload, and third-party liability each respond to a different piece of the loss — which is exactly why they need to be placed as one coordinated program.
Illustrative examples only, not guarantees of coverage. Terms vary by policy; we confirm the specifics of your program in writing.
We reach the aviation, casualty, product, and technology markets — including excess and surplus lines, Lloyd's of London, and Bermuda — that understand autonomous hardware and the way it is engineered, tested, and deployed.
Hardware, autonomy software, and field operations placed as one coordinated program, so no exposure falls into the gap between policies — which is precisely where these claims tend to surface.
We structure liability programs that anticipate disputes among manufacturer, operator, and software developer, so a contested claim doesn't leave you exposed.
We present your safety case, testing data, standards compliance, and human-oversight controls the way underwriters need to see them — which is how thin-loss-history risks still get written on clean terms.
Our embedded coverage counsel manages the claim from first notice through maximum recovery — and in a field still setting legal precedent, that expertise is the coverage.
Whether you build the hardware, write the autonomy, operate the fleet, or all three — we'll structure a program around your real exposure and respond within one business day.