OSHA has no rulebook for humanoid robots. Its robotics page says plainly that there are currently no specific OSHA standards for the robotics industry. That doesn’t mean a walking robot in your warehouse is unregulated. It means that when something goes wrong, the answer to “who pays?” will come from general workplace law, product liability, your contracts and the fine print in your insurance policies.
That question is getting practical fast. As humanoids move from demo stages into shared aisles and factory cells, buyers are running into two problems that a standards checklist doesn’t solve: who’s liable if someone gets hurt, and which insurance policy actually responds. This is a procurement briefing on how those pieces fit together for humanoid robot liability and insurance in the US in 2026. It isn’t legal advice, and your broker and counsel should see any pilot contract before you sign.
No robot rulebook, but the employer’s duty doesn’t go away
Without a robot-specific standard, employers still carry their general-industry duties. They have to recognize hazards, train workers, control hazardous energy (the lockout/tagout rules), guard machines where required, and deal with recognized serious hazards under the General Duty Clause.
OSHA has long flagged a pattern with industrial robots that applies just as much to humanoids. Many injuries happen not during normal operation but during programming, maintenance, setup and other non-routine work inside the robot’s reach. Those are exactly the moments when a pilot team is most likely to be improvising.
The technical standards that vendors and integrators map onto (ISO 10218, ISO/TS 15066, control-system performance, mobile-robot guidance) are covered in our ISO and safety standards checklist. Liability and insurance sit on top of that technical file, not in place of it.
Five parties could end up in the liability chain
| Party | Typical exposure | The question to ask |
|---|---|---|
| Employer / site operator | Workplace injury, unsafe practices, inadequate training or procedures | Is our safety case site-specific and documented? |
| Robot manufacturer | Product defect, inadequate warnings, software or firmware behavior | What do the warranty and limitation of liability say? |
| Integrator / reseller | Cell design, guarding, PLC and safety I/O, commissioning | Who’s the integrator of record in the contract? |
| Software / model provider | Policy updates that change robot behavior after acceptance | Is there change-control language for AI updates? |
| Teleoperation provider | Remote operator error, connectivity failure | How is teleoperation risk-assessed and insured? |
The last two rows are what make humanoids different from a conventional robot arm. A robot whose behavior changes with a model update, or that a remote human sometimes drives, spreads responsibility across companies that may never have set foot in your building. Contracts try to sort this out with indemnities, insurance requirements and acceptance tests. A slick demo, especially a teleoperated one, shows capability, not a finished allocation of risk. Our guide to teleoperation vs autonomy explains why that distinction matters.
The policies to check before a pilot
Workers’ compensation usually responds first when an employee is hurt, but it doesn’t end the analysis. Product and employer-liability claims can follow. Commercial general liability (CGL) is the policy to read closely: check whether autonomous or robotic equipment is excluded or needs an endorsement, because injuries to visitors and contractors land here.
Property, inland marine and equipment coverage protects the robot itself, including in transit and at off-site demos. Humanoids are expensive to damage. Product liability becomes your problem if you remanufacture, rebrand, or integrate third-party arms or hands into a system you sell. Cyber and tech E&O coverage matters for connected fleets, remote updates, and camera and microphone data; see cybersecurity risks of connected robots. Integrators need professional liability for design and commissioning errors.
Insurers and brokers are still writing specialty wordings for mobile, AI-driven robots. Expect underwriters to ask for your risk assessment, stop categories, supervision model, update policy and incident logs. That’s the same evidence a competent safety manager should be demanding anyway.
Contract language worth adding to the RFP
- Name who’s responsible for the site risk assessment, training, and lockout/tagout.
- Require certificates of insurance with minimum limits, and additional-insured status where it makes sense.
- Require notice before any safety-relevant software or model update, and define how to roll it back.
- Treat teleoperation modes separately in both the safety case and the insurance disclosures.
- Ask how near-misses and contact events are logged and shared with you.
The open question is who blinks first on AI updates. Unless the contract says otherwise, a vendor’s model update can change how a robot behaves in your building after you’ve signed off on it. Expect the first serious fights over humanoid contracts to be about that clause, not about the hardware.
Frequently asked questions
Does OSHA certify humanoid robots?
No. OSHA doesn’t issue product certificates for humanoids; it enforces workplace safety duties. Product conformity still runs through the applicable standards, declarations and, where used, third-party certificates.
If a humanoid injures a worker, whose insurance pays?
Workers’ compensation often responds first for employees, but claims against the manufacturer, integrator and others may follow. Coverage depends on the facts, the contracts and the policy wording, so get counsel and your broker involved before scaling up a pilot.
Is ISO compliance enough for insurance?
It helps with underwriting and risk control, but it doesn’t replace site procedures and training, or confirming that your CGL and other policies actually cover autonomous equipment.
Sources
- OSHA, Robotics overview: osha.gov/robotics (no specific OSHA robotics standard; hazard recognition notes)
- ISO 10218 / ISO/TS 15066 and related national adoptions (see our standards article)
- Broker and carrier briefings on AI and robotics liability (industry commentary, not law)
Last updated: October 7, 2026. To report an error, see our corrections page. Articles are drafted with AI assistance and reviewed and edited by an editor; see our editorial policy.
