Humanoid Robots and Kids: Safety, Attachment, and Chores

A 66-pound robot that can lift 154 pounds is standing in your kitchen, folding towels, while your seven-year-old watches from the doorway. Is that safe? The honest answer on humanoid robots kids safety is: physically, mostly yes if you follow a few rules; psychologically, we have twenty years of human-robot interaction (HRI) research that says children treat these machines as something between a friend and a toaster, and that ambiguity is where the real risk sits. This article walks through what the studies found, what the first shipping home humanoids actually do, and a risk-to-mitigation table you can tape to the fridge.

Key Takeaways

• 1X’s NEO weighs 66 pounds, lifts up to 154 pounds, and sells for $20,000 or $499 per month; for tasks it can’t do alone, a remote “1X Expert” can take over through its cameras at scheduled times.
• In a 2018 Science Robotics experiment, 7- to 9-year-olds went along with wrong answers from a group of robots at rates adults did not. Over-trust is the top psychological risk.
• A 2012 Developmental Psychology study found most 9- to 15-year-olds believed a humanoid had mental states and could be a friend, while still denying it civil rights. Kids build a new category, not a mistake.
• The physical hazards (pinch, crush, fall) are engineering problems with known fixes: soft shells, no-go zones, and a “one adult, arm’s length” rule.
• No humanoid on sale in 2026 is a caregiver. Robot nannies remain an ethics debate, not a product.

What is shipping to homes in 2026, and what it can actually do

A home humanoid robot is a two-legged, two-armed machine roughly the height of an adult that runs a learned control policy for household tasks and, in every 2026 product, falls back to a human operator for tasks it hasn’t mastered. The first one families can order is NEO from 1X. Its product page lists a 66-pound body, 22 degrees of freedom in each hand, a 154-pound lifting capacity, a “soft body with custom lattice polymer structure” it describes as pinch-proof, and a 22 dB noise level. Engadget reported on October 29, 2025 that NEO costs $20,000 up front or $499 a month with a $200 deposit, with deliveries starting in 2026.

The line every parent should read twice is on 1X’s own page: “For complex tasks NEO doesn’t know, an Expert from 1X can remotely supervise its actions at scheduled times to help it learn new abilities and get the job done.” That means a human employee, wearing a VR headset, sees your home through the robot’s cameras during those sessions. 1X CEO Bernt Børnich told Engadget the company blurs people so operators don’t see residents, lets owners set no-go zones, and requires owner approval before each session. He also said, plainly: “If we don’t have your data, we can’t make the product better.”

Elsewhere, the picture is thinner than the headlines. Unitree’s G1 is listed by industry guides at roughly $13,500 for a base research configuration, but it is a platform for developers, not a chore robot. Figure 03 works at BMW’s Spartanburg plant. Tesla’s Optimus, according to a roboticscenter.ai availability review updated in April 2026, has “no external sales channel, pre-order system, or waitlist,” and the same review states flatly that “no humanoid robot is ready for consumer home use in 2026.”

So the realistic 2026 scenario is: one humanoid, one household, chores only, with occasional remote-human control. Plan for that, not for the movie version.

What twenty years of child-robot research actually found

Human-robot interaction research on children is small, mostly short-term, and mostly conducted with robots far less capable than NEO. That matters. But four findings replicate often enough to plan around.

Kids conform to robots. Vollmer, Read, Trippas, and Belpaeme (2018, Science Robotics) ran the classic Asch conformity test with a group of humanoid robots giving deliberately wrong answers. Adults resisted. Children aged 7 to 9 went along with the robots’ wrong answers. The authors warned that “the potential for misuse and the potential impact of erroneous performance cannot be ignored” with vulnerable groups. A home robot that says “the stove is off” is, to a second-grader, an authority.

Kids assign robots a mind, but not rights. Kahn and colleagues (2012, Developmental Psychology) had 90 children aged 9, 12, and 15 interact with Robovie, then watched an experimenter shut the protesting robot in a closet. Most children believed Robovie had mental states and could be a friend, and thought it deserved fair treatment and protection from psychological harm, but not liberty or civil rights. Older teens attributed fewer of these qualities. Children aren’t confused; they’re inventing a category adults don’t have.

Attachment forms fast and fades when the robot gets boring. Tanaka, Cicourel, and Movellan (2007, PNAS) placed a social robot in a toddler classroom for more than five months. Interaction quality “improved steadily for 27 sessions, quickly deteriorated for 15 sessions when the robot was reprogrammed to behave in a predictable manner,” and recovered when the full behavior came back. By the end, toddlers treated it as a peer. Novelty is not the same as bond, and predictability kills engagement.

Robots teach some things well, and nobody has studied the home. Belpaeme, Kennedy, Ramachandran, Scassellati, and Tanaka’s review (2018, Science Robotics) found social robots can produce learning gains comparable to human tutoring on narrow tasks, with big caveats about short studies, small samples, and the absence of long-term data. There is no peer-reviewed study of a child living with a chore-capable humanoid. The evidence is thin here, and any company implying otherwise is guessing.

The ethical warning predates all of it. Sharkey and Sharkey (2010, Interaction Studies) argued that childcare robots raise questions about “human rights, privacy, robot use of restraint, deception of children and accountability,” with the sharpest concern being children’s emotional wellbeing if robots replace carers. Fifteen years later, nothing on sale claims to babysit, and that is the right outcome.

A risk-to-mitigation table for a humanoid in a house with kids

Risk	What the evidence says	Mitigation at home
Crush, pinch, or fall injury	NEO weighs 66 lb and lifts 154 lb; 1X claims a pinch-proof soft shell. No independent child-injury data exists yet.	”One adult, arm’s length”: no child within reach while the robot handles loads or stairs. Set no-go zones for kids’ rooms.
Over-trust of the robot’s statements	7- to 9-year-olds conformed to wrong robot answers (Vollmer et al., 2018).	House rule: the robot never gives permission. “Ask the robot, check with a parent.” Practice catching it being wrong.
Attachment and category confusion	Most 9- to 15-year-olds saw Robovie as a friend with feelings (Kahn et al., 2012).	Use “it,” not “he.” Name it after an appliance, not a person. Talk about it as a tool the family owns.
A remote stranger sees inside your home	1X Experts view through cameras in scheduled sessions; blurring and approval are owner-controlled (Engadget, 2025).	Schedule Expert sessions when kids are at school. Turn on blurring. Put bedrooms and bathrooms in no-go zones.
Robot as substitute caregiver	Sharkey & Sharkey (2010): the core ethical hazard is emotional, not mechanical.	The robot is never “watching” a child. It does chores while an adult is home. Full stop.
Unsafe conversational answers	PIRG’s 2025 AI-toy tests found a plush toy explaining how to strike matches (The Register, Nov 13, 2025).	Treat the robot’s built-in chat like any chatbot: test it yourself with the questions a kid would ask before the kid does.
Novelty crash and disappointment	Toddler engagement collapsed when the robot became predictable (Tanaka et al., 2007).	Don’t buy it “for the kids.” Buy it for laundry. Expect interest to fade within weeks.

What to actually do if a humanoid enters your home

Run the “robot is wrong” drill in week one

Because the conformity result is the best-replicated child-specific finding, spend the first week deliberately catching the robot in errors. Ask it something you know it will get wrong or can’t know, then let your kid see you check. The goal is a reflex: robot says X, human verifies X. A useful phrase for younger kids is “the robot guesses, people check.”

Write the no-go zones with your kid, not for them

Most home robots, including NEO, let owners mark areas the robot (and any remote operator) cannot enter. Draw the floor plan together and let your child mark their own room. This turns a privacy setting into a lesson about consent and property. It also gives the child a sense of control that the research on attachment suggests matters more than we assume.

Schedule remote-operator sessions like a contractor visit

Treat a 1X Expert session the way you’d treat a plumber: known time, adults present, kids elsewhere, blurring on. Tell your kids explicitly that “during the green-light hours a person at the company can see through the robot’s eyes.” Older kids will ask good questions about where the video goes. 1X says limited sensor data from autonomous requests is processed but not stored and that sharing data for improvement is optional; verify the current policy in the app before you sign.

Keep the language boring on purpose

“It,” “the robot,” “the machine.” Not “he,” not a human name, not “he’s tired.” Kahn’s 2012 data shows kids will grant mental states anyway, so the point isn’t to prevent it; it’s to keep the adults from modeling it. The child’s category-building will happen regardless. Yours shouldn’t feed it.

What not to do

Don’t use the robot as a bargaining chip (“the robot will be sad if you don’t clean up”). Don’t leave a child under 10 alone in a room with an operating humanoid, even briefly. And don’t let the robot be the thing that answers “can I have a snack?” It has no authority in your house, and your kid needs to hear you say that out loud.

What to Watch For Over the Next 3 Months

• Week 4: Your kid has stopped narrating the robot’s every move. Good. If they’ve started confiding in it or asking it for permission, revisit the “robot guesses, people check” rule.
• Month 2 red flags: The child avoids rooms when the robot is active (fear), or insists the robot “likes” them and gets upset when it’s powered down (attachment). Neither is an emergency, but both deserve a conversation.
• Month 3 self-check: Have you ever left the room assuming the robot “had it”? If yes, that’s the Sharkey problem arriving quietly. Reset the rule.

Frequently Asked Questions

Can a home humanoid robot physically hurt a child?

Any 66-pound machine that lifts 154 pounds can hurt someone. 1X built NEO with a soft lattice shell it calls pinch-proof, and it operates at whisper volume, but there is no independent child-injury data for any home humanoid. Keep kids at arm’s length during tasks and mark kids’ rooms as no-go zones.

Will my child get emotionally attached to a robot?

Probably, and quickly. Kahn et al. (2012) found most 9- to 15-year-olds believed a humanoid had feelings and could be a friend. Tanaka et al. (2007) saw toddlers treat a robot as a peer within months. Attachment isn’t harmful by itself; the risk is when it displaces people or becomes a bargaining chip.

Is someone really watching through the robot’s cameras?

Sometimes, yes. 1X’s NEO uses remote “Experts” for tasks it can’t do alone, in sessions the owner schedules and approves. 1X says it can blur people and honor no-go zones. Schedule those sessions when kids are out, and turn every privacy option on.

Can I use a humanoid robot as a babysitter?

No product on sale in 2026 claims to be one, and Sharkey and Sharkey’s 2010 ethical review is still the reference for why that’s the right line. The robots do chores. An adult supervises children. Those two facts shouldn’t blur.

Should I explain to my kid that the robot isn’t alive?

Yes, and expect it not to fully land. Research shows kids build a middle category, “alive-ish,” rather than accepting either answer. Keep your own language mechanical (“it,” “the machine”) and use the robot’s mistakes as teaching moments rather than lecturing.

Are these robots worth $20,000 for a family?

That’s a household budget question, not a safety one. What the evidence supports: the chores work is narrow, the remote-human fallback is real, and the child-development upside is unproven. Buy it for laundry if the math works. Don’t buy it for the kids.

---

About the author

Ricky Flores is the founder of HiWave Makers and an electrical engineer with 15+ years of experience building consumer technology at Apple, Samsung, and Texas Instruments. He writes about how kids learn to build, think, and create in a tech-saturated world. Read more at hiwavemakers.com.

---

Sources

1. Vollmer, A.-L., Read, R., Trippas, D., & Belpaeme, T. (2018). “Children conform, adults resist: A robot group induced peer pressure on normative social conformity.” Science Robotics, 3(21). https://doi.org/10.1126/scirobotics.aat7111
2. Kahn, P. H., Jr., Kanda, T., Ishiguro, H., et al. (2012). “‘Robovie, you’ll have to go into the closet now’: Children’s social and moral relationships with a humanoid robot.” Developmental Psychology, 48(2). https://doi.org/10.1037/a0027033
3. Tanaka, F., Cicourel, A., & Movellan, J. R. (2007). “Socialization between toddlers and robots at an early childhood education center.” PNAS, 104(46). https://doi.org/10.1073/pnas.0707769104
4. Belpaeme, T., Kennedy, J., Ramachandran, A., Scassellati, B., & Tanaka, F. (2018). “Social robots for education: A review.” Science Robotics, 3(21). https://doi.org/10.1126/scirobotics.aat5954
5. Sharkey, N., & Sharkey, A. (2010). “The crying shame of robot nannies: An ethical appraisal.” Interaction Studies, 11(2), 161–190. https://benjamins.com/catalog/is.11.2.01sha
6. 1X Technologies. (2026). “NEO Home Robot.” https://www.1x.tech/neo
7. Engadget. (Oct 29, 2025). “1X Neo is a $20,000 home robot that will learn chores via teleoperation.” https://www.engadget.com/ai/1x-neo-is-a-20000-home-robot-that-will-learn-chores-via-teleoperation-040252200.html
8. RoboticsCenter.ai. (2026). “Tesla Optimus price and availability.” https://www.roboticscenter.ai/blog/tesla-optimus-price-availability
9. There’s a Robot for That. (2026). “Humanoid robot companies 2026: complete guide.” https://theresarobotforthat.com/blog/humanoid-robot-companies-2026-complete-guide/
10. The Register. (Nov 13, 2025). “AI-enabled toys teach kids how to play with fire, sharp objects” (on U.S. PIRG Education Fund’s Trouble in Toyland 2025). https://www.theregister.com/2025/11/13/ai_toys_fmatches_knives_kink/

Related reading: what current robots actually can and can’t do, the real math on a $499-a-month home robot, and should you let a robot babysit.