Market
Funding, raised vs shipped, investors and safety.
Safety records
Public reports about robots, from regulators and the press. Each record links to its source.
- Records
- 6,008
- Registers
- 14
- Earliest record
- 1987
What this tells you
98% of records
come from hospitals and roads.
That reflects which governments publish robot reports, not where robots fail.
95 records
cover all factories, workplaces and consumer products.
A thin record is a gap in what is published, not proof of safety. Ask for incident logs.
4 records
from China, home to 325 robot makers.
Absence is not safety, for any country.
Where it happened
The setting matters: a hospital is not a home.
Surgical robots · FDA MAUDE, Health Canada
- Public roads →1,303
Driverless-vehicle crash reports · NHTSA
Workplace accident records · OSHA, Japan MHLW, Taiwan MOL
Recalls and product alerts · EU Safety Gate, CPSC, Health Canada, UK OPSS, AU Product Safety and 3 more
News reports checked one at a time · Press reports
Where it was recorded
By the country of the registry. Few records can mean few registers, not few incidents.
- United States5,944
4 registers
- Press, many countries24
News reports
- Canada12
Health Canada
- European Union11
EU Safety Gate
- Japan6
2 registers
- China4
2 registers
- Australia3
AU Product Safety
- United Kingdom3
UK OPSS
- Taiwan1
Taiwan MOL
How serious
- Death reported
Reports that mention a death, by source:
325 US FDA (medical devices)31 US OSHA (workplaces)2 press reports2 Japan health and labour ministry1 US NHTSA (road vehicles)1 Taiwan labour ministry - Serious injury41
- Minor injury121
- Recall26
- Property damage only1,112
- No injury reported62
- Not graded by the source4,284
A report is not a verdict.
Only recalls and finished workplace investigations say what caused it.
No league table.
More robots on the road means more reports. Rates need miles or hours, and few makers publish them.
Browse all 6,008 records, filtered by registry, severity, country or year. If we have a record wrong, use the form on the company's page. How we handle corrections
The registers and the count-versus-rate chart
Where the records come from
- FDA adverse event reportsFDA4,577325 death reportsReports involving surgical robots. A report is an allegation, not a finding.2000–2026
- NHTSA Standing General OrderNHTSA1,3031 death reportCrash reports that driverless-vehicle operators are legally required to file.2025–2026
- OSHA accident investigationsOSHA6031 death reportsUS workplace accidents where an investigator found a robot in the chain of events.1987–2024
- Press and public records242 death reportsIndividually checked reports from news and regulator announcements.2008–2026
- Health Canada recallsHealth Canada122019–2026
- EU Safety Gate alertsEU Safety Gate112010–2025
- Japan MHLW workplace recordsMHLW52 death reports
- CPSC recallsCPSC42022–2026
- UK OPSS recallsOPSS32022–2025
- Product Safety AustraliaACCC Product Safety32023–2026
- China DPRC consumer recallsSAMR Defective Product Recall Technical Centre22025–2026
- China SAMR recallsSAMR (State Administration for Market Regulation)22021–2024
- Japan NITE product accidentsNITE12021–2021
- Taiwan Ministry of LaborTaiwan MOL Occupational Safety and Health Administration11 death report2023–2023
Why there is no league table
A crash count means little without miles driven. Ranked by rate, the order flips.
6.0×lower reporting rate for the operator with the most reports
Miles as of 31 Mar 2026 for Waymo and 24 Mar 2026 for Zoox. Counting reports ranks these two in the opposite order to measuring them.