Evidence
The first humanoid safety cases are lawsuits and test-rig videos, not injuries
| Date | Case | Country | What happened | Source |
|---|---|---|---|---|
| May 2025 | Unitree H1 test-rig malfunction | China | A suspended humanoid flailed its arms at two engineers; no injuries | Source |
| Sep 19, 2025 | Hinterdobler v. Tesla and FANUC | United States | Technician says a robot arm knocked him unconscious; seeks $51 million | Source |
| Nov 21, 2025 | Gruendel v. Figure AI | United States | Former safety head alleges robots could fracture a skull; Figure denies | Source |
| Jul 2024 | OSHA severe injury report study | United States | 77 robot accidents and 93 injuries, 2015 to 2022 | Source |
Registry
Four registry incidents involve physical harm
Registry incidents meet a stricter bar than the cases above: public sources name the developer, the failed control and who found it.
Exposure
10 of 17 agent types rate very significant or critical on physical access or behavioral and policy boundary
From the attack surface ratings, the agent types where this risk has the widest path in.
| Agent type | Dimension | Rating |
|---|---|---|
| Coding and research agents | Behavioral and policy boundary | Critical |
| Workflow agents | Behavioral and policy boundary | Critical |
| Commerce agents | Behavioral and policy boundary | Critical |
| Humanoid robots | Behavioral and policy boundary | Very significant |
| Industrial robots and cobots | Physical access | Very significant |
| Autonomous trucks | Behavioral and policy boundary | Very significant |
| Drones | Behavioral and policy boundary | Very significant |
| Fleets and swarms | Behavioral and policy boundary | Very significant |
| Medical devices | Behavioral and policy boundary | Very significant |
| Enterprise agents | Behavioral and policy boundary | Very significant |
Controls that bound it
- Behavioral Envelopes: hard limits on action enforced outside the model
- Human Oversight: human authority to approve, intervene and stop
- OT and ICS Integration: how agents connect to industrial control systems
Rules
Robot safety standards were rewritten in 2025; China added humanoid standards in 2026
| Rule | Jurisdiction | What it requires | Status |
|---|---|---|---|
| ISO 10218-1 and -2:2025 | International | Industrial robot safety, with collaborative and cybersecurity rules | Published 2025 |
| Humanoid robot standards system (TC08) | China | Lifecycle safety and ethics among six areas | Released Feb 28, 2026 |
| EU Machinery Regulation 2023/1230 | EU | Covers AI-driven machines | Applies Jan 20, 2027 |
Status colors: green in force, blue voluntary standard or framework, amber adopted and phasing in, gray draft or pending, red withdrawn.
Record
Page Record
- Risk group
- Human Risks
- Datasets
- ai_risk_evidence.csv (4 rows); registry incidents (4 linked); ai_attack_surface.csv
- Key sources
- Unitree H1 test-rig malfunction
Hinterdobler v. Tesla and FANUC
Gruendel v. Figure AI
OSHA severe injury report study - Last verified
- October 2026