Meta built a humanoid with Fourier? What the report really says

Meta may already have a humanoid prototype. According to original reporting by The Information, it built one with China's Fourier and is working on another version. The story suggests that major AI platforms want to move from software and wearables into physical household tasks. One caveat matters: as of 1 November 2026, the alleged Meta machine has not been publicly demonstrated or independently verified.

What emerged on 31 October?

The Information cited people familiar with the project to report a Meta–Fourier humanoid prototype and development of a successor. Reuters covered the account that day but explicitly said it could not independently confirm it. That leaves one original anonymously sourced report, not two independent manufacturer confirmations.

The report describes a possible physical embodiment of Meta's Muse AI agent, potentially aimed at household tasks. It does not establish a product launch. No price, shipment timetable, task-success data or degree of autonomy has been published for the alleged machine.

Reported facts versus missing evidence

Why Fourier would be a notable partner

Fourier builds humanoids and rehabilitation systems. Its existing GR-3 platform, profiled in RoboMorrow's database, prioritises human interaction. Official documentation lists approximately 165 cm in height, 71 kg of mass, up to 55 degrees of freedom and VR or exoskeleton teleoperation. Those are GR-3 specifications, not specifications for Meta's alleged robot. We do not know whether the reported prototype shares its hardware.

This distinction also applies to photographs. A convincing image of Fourier's current product is not evidence of what an unannounced Meta robot looks like.

An AI assistant is not yet a useful household robot

Physical tasks require much more than conversation. A home robot must handle unfamiliar objects, move safely around people and animals, recover from failed grasps and know when to seek help. A capable language model cannot eliminate mechanical, perception and safety constraints.

Before evaluating a consumer launch, we would need repeatable task trials, successful-operation rates, teleoperator interventions, error records and measured battery runtime. The other essential question is the total cost of making a humanoid useful, beyond the hardware price.

US–China politics add another unanswered question

A connected mobile robot raises questions about manufacturing origin, camera and microphone data, updates, emergency controls and market authorization. We have previously covered new US restrictions potentially affecting some foreign robots. The policy does not establish that this unannounced Meta project would be covered, barred or sold in any specific country.

Five things to verify next

• An on-record acknowledgement from Meta or Fourier.
• Actual prototype footage with human teleoperation disclosed.
• Repeatable home trials with success and intervention rates.
• Safety, privacy and stop-control documentation.
• Commercial plans covering supported markets, service, pricing and deliveries.

RoboMorrow verdict

Potentially huge news, but still limited verification. A Meta AI agent in a mobile humanoid could create a new competitive front in home technology. Today it remains reporting about a prototype the public has not seen. It is neither a finished product announcement nor evidence of reliable domestic autonomy.

Sources and method

• The Information — original report, 31 October 2026 (subscription required).
• Reuters — account explicitly not independently verified, 31 October 2026.
• Fourier — official GR-3 documentation, for background only.

Redakcja RoboMorrow · checked 1 November 2026 · source-based reporting, no prototype access or hands-on testing. No affiliate link: no verified retail offer.