# Which home robot is the safest?

**Which Home Robot Is the Safest? (2026)** · Last verified Jul 26, 2026. 4 sources.

**There is no safest home robot the record can crown. None has published child-safety or home-safety testing, and the one hard fact is that only the Unitree G1 carries safety incidents on our record.** We will not tell you which of these robots is safe, because none has published the testing that would let anyone say so. Not one has released child-safety results, a home-safety certification, or force limits for a robot around your family. A published international standard for household robots exists, ISO 13482, and there is no evidence any of these machines has been tested to it. What the verified record does hold is an incident count, and it is lopsided: the Unitree G1 carries the field's safety incidents, while the Optimus, the NEO, the Figure 03, and the Apollo each show zero. Read that gap honestly. Five is a real signal for the G1. Zero for the others is mostly the absence of deployment, not a proven safety case, because those machines are barely in homes at all. The 1X NEO adds a wrinkle: it is teleoperation-assisted, so safety partly depends on a remote human. The honest verdict is that safety here is undisclosed across the board, with one machine already showing incidents. We crown no winner. This is general information, not safety advice.

## Quick answers

- **Has any maker published home-safety testing?**: None has
- **Which has incidents on the record?**: Only the Unitree G1 (5)
- **Does a zero record mean it is safe?**: No, mostly not-yet-deployed
- **Is there a safety standard for home robots?**: ISO 13482 exists

## Short answer

Parents and first buyers want a safest pick, and the honest answer is that no one has earned the label. None of these makers has published child-safety testing, a force limit, or a home-safety certification, and there is no sign any has been tested to [the household-robot safety standard, ISO 13482](/home-robots/optimus-child-safety.md). The only hard safety datum on the record is the incident count, and it points one way: the Unitree G1 carries the field's incidents, while the others sit at zero. That zero is easy to misread. For the Optimus, the Figure 03, and the Apollo, it mostly means they are not in homes yet, not that they are proven gentle. The 1X NEO leans on a remote operator, which moves the safety question rather than settling it. For the deep answers about a home humanoid around children and pets, the Optimus safety family covers each worry one at a time. We report the record and never fill a safety silence with reassurance. This is general information, not safety advice.

## What actually exists today

| What people ask about | Status | The reality |
| --- | --- | --- |
| Published home-safety testing | None | No maker has published child-safety or home-safety testing, force limits, or a home-safety certification. A household-robot standard exists (ISO 13482); there is no evidence any of these was tested to it. |
| Safety incidents on the record | Only the Unitree G1 | The verified record holds safety incidents only for the Unitree G1; the Optimus, NEO, Figure 03, and Apollo each show zero. The live count on this page is resolved from the record. |
| What a zero record means | Not-yet-deployed, not proven safe | For the home-aimed machines, zero mostly reflects that they are not deployed in homes at scale, not that they are proven safe. Absence of incidents is not a safety case. |
| A remote human in the loop | 1X NEO | NEO ships teleoperation-assisted, so its behavior can depend on a remote operator. That can help in a hard moment and adds a dependency; either way it moves the safety question rather than settling it. |
| A safety standard to test against | ISO 13482 exists | ISO 13482 is a published international safety standard for personal care robots. It is real and checkable as the category bar; whether any machine here is certified to it is undisclosed. |

## How to read the safety question

A safety comparison lives or dies on not over-reading a clean record. Hold each figure at its true weight: an incident count is real, and so is the reason a count might be zero.

- **The real signal: The G1's incidents are on the record**: The Unitree G1 carries the field's safety incidents, more than any peer here. That is a genuine, verified signal for a machine people can actually buy today, and it deserves weight.
- **The misread: Zero is mostly not-yet-deployed**: For the Optimus, Figure 03, and Apollo, a zero record largely reflects that they are barely in homes, not that they are proven gentle. Do not read an absence of incidents as a safety case.
- **The wrinkle: Teleoperation cuts both ways**: The NEO's remote operator can intervene in a hard moment and is also a dependency and an unknown. It changes the shape of the safety question rather than answering it.
- **Where it stands: No winner, and no reassurance**: With no published testing anywhere, there is no safest home robot to crown, and we will not write that any is safe. When a maker publishes real safety testing, this page gets a dated update.

## What would change this

We track the whole field against the verified records and [the live incident record](/robot-recalls-and-safety-incidents.md), so you do not have to. Any one of these moves the answer, and each becomes a dated line the week it happens:

- Any maker publishes child-safety or home-safety testing, a force limit, or an ISO 13482 certification.
- A new verified safety incident lands on any machine, or an existing one is resolved or retracted.
- An independent assessment measures a machine's contact force, fall behavior, or stair handling.
- Safety scrutiny of home robots follows real incidents. Right now the verified record holds5 safety incidentsacross the home robots we compare here, essentially all of them on the Unitree G1, and every new one is logged the week it lands.

## How we know this

The incident counts are the non-retracted safety incidents the verified record holds for each machine, resolved live on this page. The safety standard is a published ISO document. Every absence is stated plainly. We present no safety result no maker has published, and we never write that a robot is safe.

- [DEPLOY verified incident record (the live counts)](/robot-recalls-and-safety-incidents.md) (live)
- [Is a Tesla Optimus safe around children and pets (the safety family)](/home-robots/optimus-child-safety.md) (live)
- [Unitree G1: the verified registry record (the machine with incidents)](https://registry.deploy.report/models/unitree-g1) (live)
- [ISO 13482:2014, safety requirements for personal care robots](https://www.iso.org/standard/53820.html) (in force)

If any maker publishes home-safety testing, or a new incident lands, or you can show us we got something wrong, our corrections process logs the response on the record.

## Common questions

### Which home robot is the safest?

There is no verified safest home robot. None of the makers has published child-safety or home-safety testing, and there is no evidence any is certified to the household-robot standard ISO 13482. The only hard fact on the record is that the Unitree G1 carries the field's safety incidents, while the others show zero, and that zero mostly reflects that those machines are not in homes yet, not that they are proven safe. We crown no winner. This is general information, not safety advice.

### Does a zero incident record mean a robot is safe?

No. For the home-aimed humanoids, a zero record mostly means they are barely deployed in homes, so there is little opportunity for an incident to be recorded. It is an absence of data, not a demonstrated safety case. A machine with a clean record and no published testing is unproven, not proven safe.

### Why does the Unitree G1 have incidents when the others do not?

Partly because it is the one people can actually buy and use today, so there is real-world activity to record. The verified record holds the field's safety incidents on the G1. That is a genuine signal worth weighing, and it is also a reminder that the zeros elsewhere reflect a lack of deployment rather than proof of safety.

### Is a teleoperated robot like the 1X NEO safer?

It is different, not clearly safer. A remote operator can step in during a hard moment, which can help, but it is also a dependency and an unknown, and it raises its own privacy questions. Teleoperation moves the safety question rather than answering it, and 1X has not published safety testing for home use.

## Keep reading

- [The one you can buy: Optimus vs Unitree G1](/home-robots/optimus-vs-unitree-g1.md): Where the safety gap really sits.
- [Which home robot is most private?](/home-robots/most-private-home-robot.md): The privacy companion to this page.
- [Tesla Optimus vs the field: which home robot should you compare?](/home-robots/home-robot-comparison.md): Every home-robot comparison, grounded in the verified record, one page.
- [Which home robot is the best value?](/home-robots/best-value-home-robot.md): Best for you

_Canonical URL: /home-robots/safest-home-robot. HTML version: [link](/home-robots/safest-home-robot)._