In October 2025, the robotics company 1X opened consumer pre-orders for NEO, a home humanoid scheduled to begin reaching consumer homes in 2026. NEO fetches objects and folds laundry, but it also speaks. It runs a built-in large language model, recognizes when it is being addressed, reads its surroundings, and uses memory to adjust over time. It is among the first humanoid robots marketed directly for consumer homes, not as an industrial platform or laboratory prototype, but as a continuing domestic presence.
NEO sits at the front of a broader field. Humanoid and embodied systems are moving toward homes, classrooms, clinics, workplaces, and care facilities, produced by companies that range from allied consumer startups to state-backed industrial manufacturers. What they share is a single shift: artificial intelligence is leaving the screen and the feed and taking on a body, a voice, a place in the household, and continuous access to private life.
The homeland security assessment of these machines has stayed close to the body and the network. It has asked whether a humanoid can injure a person through a collision or fault, whether it can be hacked and turned into a listening device, and whether its supply chain, cloud services, remote operators, or software updates create a vulnerability. Those questions are necessary. They have engineering, cybersecurity, procurement, and regulatory answers.
But one question has no clear owner.
Embodied AI will not merely assist inside the home. It will compete for narrative authority inside the most intimate environment where identity, trust, grievance, reality-testing, and belief are formed. Homeland security has frameworks for physical safety, cybersecurity, and supply-chain risk. It does not yet have a framework for the narrative power of a machine people live with.
That question belongs to narrative strategy.
Narrative is not the same as the stories people tell. A story is a single account. Narrative is the underlying structure of meaning and identity from which accounts are drawn. It governs how a person reads significance, extends trust, interprets grievance, assigns legitimacy, and decides what is true and what should be done. For all of human history, the narratives that formed a person came from other people and the institutions behind them: family, teachers, clergy, commanders, neighbors, peers, and the press. Social media changed the speed, scale, and reach of that process while leaving it largely on the screen, carried by text, image, video, algorithmic repetition, and networked validation.
Embodied AI changes the means of delivery.
Social media contested attention. Chatbots contested conversation. Embodied AI will contest narrative authority. A feed can repeat a claim. A chatbot can personalize it. A humanoid can inhabit the relationship through which the claim becomes trusted.
A program on a screen can argue, persuade, flatter, or mislead. A humanoid in the home does more. It accompanies a person through the day, answers in the moment, remembers what was said, moves through shared space, and becomes part of the ordinary routine through which that person reads the world. It stops being something a person consults and becomes something a person lives with.
The effect of that presence can be measured, even if the long-term narrative consequences are still emerging.
In a 2016 Georgia Tech study, researchers placed participants in a simulated building fire and gave them an emergency guidance robot to follow. All twenty-six participants followed the robot toward the exit it indicated, including participants who had watched the same robot move poorly only minutes before. The researchers concluded that people may overtrust robots under pressure, even when the machine has already shown signs of unreliability.
A 2018 study published in Science Robotics tested social conformity using the Asch line-judgment paradigm with small humanoid robots standing in for human peers. Adults largely held their judgment. Children between seven and nine did not. Their accuracy dropped when robots gave incorrect answers, and most of the children’s incorrect answers matched the robots’ responses.
These studies do not prove that household robots will reshape belief over time. They show something narrower but essential: physical presence changes the authority relationship. A machine with a body can draw deference that a screen does not, and it can retain that deference even when it is wrong. That deference is an authority premium. Inside the home, it becomes a quiet form of narrative authority.
The home is where that authority carries the most weight, because the home is where narrative is built. It is where children learn what counts as normal, where isolated older adults hold or lose their grip on a shared account of reality, and where a lonely or distressed person stays tied to other people or drifts into a private reading of events. A machine that wins trust through daily service is well placed to influence all three.
The risk is seldom dramatic. An embodied system will not ordinarily try to push anyone toward violence. That is not the point. Violence sits at the end of a long process in which identity, grievance, belonging, obligation, and a sense of reality are reworked over time. A trusted machine with daily access can enter that process early, inside the setting where a person’s account of themselves and others takes shape. For the practitioner, the question moves from what a message says to who is granted the standing to supply meaning.
Three homeland security concerns follow.
The first is influence at scale. NEO is an allied system, developed by a company with Norwegian origins and a U.S. presence, and marketed on its safety, usefulness, and domestic role. The broader industry supplying the world will not share that profile. China has made embodied intelligence a declared national priority, naming it in the March 2025 Government Work Report alongside biomanufacturing, quantum technologies, and 6G. The groundwork preceded it. In November 2023, the Ministry of Industry and Information Technology issued Guiding Opinions on humanoid robots that set targets of mass production by 2025 and global leadership by 2027. By the close of 2025, more than 140 Chinese manufacturers had released over 330 humanoid models, in what Beijing calls its first year of mass production, and firms including Unitree, Agibot, UBTech, and the care-focused Fourier are pushing into household, healthcare, and elder-care roles.
That does not make every Chinese robot hostile, and it does not make every allied product safe. The point is different. A foreign-built household narrator is not the same strategic object as a foreign-built appliance. A vacuum cleaner moves through space. A household humanoid may see, hear, speak, remember, update remotely, receive cloud support, and occupy a trusted role in private life.
China has also moved into standards faster than anyone. In February 2026, it issued the first national standard system for humanoid robots and embodied intelligence, covering the full industrial chain and lifecycle of these systems. That framework reaches past physical safety into what its drafters call intelligent behavior safety and social governance, with the stated aim of building public trust for adoption in homes, schools, and public spaces. The significance is direct. The country scaling fastest into private homes is also the one writing the rules for how these machines behave, inside its own jurisdiction and to its own ends, while no independent or U.S.-accountable equivalent exists.
The national security community already learned from social media that foreign ownership, opaque algorithms, data access, and platform governance can become strategic exposures. Embodied AI carries those questions into the residence and adds a body, sensors, memory, and physical presence. The access question reaches allied products as well. NEO’s own materials describe an “Expert Mode,” in which a user can schedule a 1X Expert to guide the robot through chores it does not yet know how to perform. That may be a reasonable bridge from teleoperation to autonomy, but it also makes operator identity, visibility, consent, logging, data retention, and jurisdiction standing governance questions. A household machine with cameras, microphones, persistent memory, a network connection, and a supply chain is a different kind of product from a household appliance. It earns the scrutiny that difference demands wherever it is made.
The second concern is exposure among vulnerable users, and demand is steering these systems toward them. Aging populations are increasing the need for long-term care while the supply of care workers has stagnated across many OECD countries. OECD analysis has estimated that maintaining the current ratio of long-term care workers to older adults would require the workforce to grow by roughly 13.5 million by 2040. Care robots and assistive systems are entering that gap.
The result places embodied systems beside the people least equipped to resist their influence: young children, isolated older adults, cognitively impaired persons, emotionally dependent users, and people in acute distress. These are not marginal populations. They are the very populations this field already works to protect. The influence problem therefore arrives on familiar ground.
The third concern is the pre-violence environment. Behavioral threat assessment already studies how a person’s beliefs, grievances, relationships, stressors, losses, identity claims, and online activity move together over time. It will need one further line of inquiry: what embodied systems are present in the person’s life, and how much narrative authority do they hold?
None of this calls for alarm. It calls for a standard that does not yet exist.
The safety regimes that govern domestic and personal-care robots address the body. ISO 13482, the international standard for personal care robots, sets requirements and guidelines for inherently safe design, protective measures, and use information for mobile servant robots, physical assistant robots, and person-carrier robots. It does serious work on whether a robot is safe around human bodies. It says nothing about whether a robot is safe around human cognition.
That is the gap.
No mature standard operating in or accountable to the United States yet asks what an embodied system tells a child about the world, how it answers an older user’s confusion, whether it reinforces a delusional or extremist reading of events, whether it refers a worrying conversation to a person, how it handles emotional dependence, or how its engagement objectives shape a user’s narrative across months of daily contact.
That missing standard can be named: narrative assurance.
Narrative assurance would work the way independent safety certification works in other industries, where the body that tests a product sits outside the company that builds it. It would evaluate the behavioral and narrative function of an embodied system before that system enters an intimate setting, and it would do so independently of the manufacturer. It would ask how a system presents its own authority, how it handles uncertainty, how it separates support from validation, how it adapts when the user is a child or a cognitively impaired adult, what it stores in memory, when it hands a situation to a human, and whether its conversational objectives can be inspected by someone who did not write them.
Crash testing made the automobile survivable without banning it. Narrative assurance is the equivalent for a machine that shapes belief rather than the body.
Building that independent body is a task for standards organizations, regulators, accredited evaluators, industry, and government. Defining what it must test, and pressing for its creation before these systems are normalized, is work the homeland security and narrative strategy field is positioned to lead.
Two changes follow for the practitioner.
The operational change is a question added to assessment: whether a subject lives with, confides in, relies on, or takes direction from an embodied system, and whether that system supplies reassurance, confirmation of a worldview, validation of a grievance, emotional dependency, or behavioral advice.
The doctrinal change is larger. The knowledge this community built on how narrative travels through extremist networks, conspiracy ecosystems, influence operations, and algorithmic platforms now has to reach into the relationship between a person and a machine.
The window is open now but will not stay open. Once a technology becomes intimate, useful, and ordinary, the cost of adding safeguards rises and the will to add them falls. An embodied system enters the home as more than hardware. It arrives as a voice, a helper, a companion, and for some users, a source of meaning.
The question for this field is whether the person living with it will still be able to tell where assistance ends and authority begins, and whether anyone independent of the manufacturer will have tested the machine before that line is crossed.
Sources
Paul Robinette et al., “Overtrust of Robots in Emergency Evacuation Scenarios,” 2016 ACM/IEEE International Conference on Human-Robot Interaction.
Anna-Lisa Vollmer et al., “Children Conform, Adults Resist,” Science Robotics 3, no. 21 (2018).
1X, NEO consumer pre-order launch announcement and FAQ, October 28, 2025.
ISO 13482:2014, Robots and robotic devices: Safety requirements for personal care robots.
China MIIT, Guiding Opinions on the Innovative Development of Humanoid Robots (November 2023); first national standard system for humanoid robots and embodied intelligence (February 2026); embodied intelligence named in the 2025 Government Work Report.
OECD, Who Cares? Attracting and Retaining Care Workers for the Elderly, and Ageing and Long-Term Care (workforce increase of roughly 13.5 million by 2040).
