This week: the FCC's foreign-robot restrictions run into a real factory, China pushes robots into homes and shops, and a practical way to see what your robot is sending home.
Top threat: when a robot ban collides with security updates
Intel has asked the FCC for a conditional waiver so it can keep operating foreign-made automated guided vehicles (AGVs) in its US chip plants. The petition was filed on October 16 and became public on October 19 (Quartz). According to Tom's Hardware, the AGVs move wafers and materials between tools, they need ongoing software and firmware updates, and Intel says it is unclear whether those updates would stay compliant with the FCC's restrictions on foreign-made robots. The largest suppliers of these vehicles are Japanese. No FCC response has been reported.
Why it matters (analysis): this is the first big test of how the robot restrictions treat robots that are already installed. If a rule makes operators unsure whether they can apply a firmware update, the safest-looking choice on paper (freeze the fleet) is the worst one for security: known flaws stay unpatched. Humanoid buyers should ask vendors now how updates will be delivered and supported under the FCC rules. Background: FCC robot ban explained.
Policy watch
- China: on October 19, the Ministry of Commerce and eight other departments issued guidance listing robots among smart products whose supply should expand, and calling for embodied intelligence in household services and commercial logistics. The coverage we reviewed mentions no safety or security provisions. Humanoids Daily
- Home robots and privacy: with consumer humanoids heading into homes, we published a guide to what they collect and which rules apply (GDPR, the EU AI Act and FTC enforcement): Humanoid robot privacy: what home robots see, hear and share.
Tool of the week: egress monitoring for robots
Alias Robotics' assessment of the Unitree G1 found the robot sending telemetry to remote servers every five minutes without a visible indicator (arXiv 2509.14139). The practical defense is simple: put robots on their own network segment, log every outbound connection, and allow-list only the destinations you have approved. Unexpected endpoints are your first sign that a robot is sharing more than the vendor said. More checks in our privacy checklist.
Directory pick: Alias Robotics
The robot security firm behind the G1 telemetry research and several published robot vulnerability studies. Read our Alias Robotics profile or browse the full security directory.
Know someone who should read this? Forward it, or send them humanoidthreats.com/newsletter/
Get the free Humanoid Robot Security Checklist, and see every known flaw in the Humanoid Robot Vulnerability Tracker.
Get the weekly Humanoid Threats Brief → Subscribe


