← Back to stories Close-up of a computer screen displaying ChatGPT interface in a dark setting.
Photo by Matheus Bertelli on Pexels
凤凰科技 2026-03-10

JD rolls out remote deployment service for OpenClaw agents amid security warnings

JD brings "one-click" OpenClaw install to ordinary users

JD (京东) has launched a remote deployment service for the open‑source AI agent OpenClaw (开源AI智能体软件 OpenClaw), reportedly aimed at removing technical barriers for everyday users. It has been reported that buyers who purchase the service through JD’s 3C digital services flagship store can schedule a one‑on‑one remote session with a professional engineer for pre‑installation of Lenovo (联想) Baiying AI office Skills and minute‑level deployment — claims marketed as “newbies can get started in seconds.” The pitch: powerful agent automation without leaving home.

What OpenClaw does — and why some call it "the lobster"

OpenClaw has become popular because, unlike chatbots that only offer suggestions, it can take actions after obtaining permissions — sorting files, processing mail, writing code and more. Users have nicknamed it “lobster” (龙虾) because of its red icon. Proponents say this breaks the old “AI only talks” limitation; critics ask a sharper question: what happens when an agent that can act autonomously goes wrong?

Security experts urge caution

Security researchers warn of real risks. It has been reported that in the absence of strict permission controls, auditing and hardening, agents like OpenClaw can be induced by malicious prompts, misconfiguration or takeover to perform unauthorized actions, leading to data leakage or device compromise. Experts recommend closing unnecessary public access, enforcing strong identity and access controls, enabling encryption and continuous security auditing, and following official security bulletins closely.

Regulatory backdrop and what to watch next

As agent software spreads, regulators in China and elsewhere are sharpening scrutiny of data governance and AI security. Will convenience trump caution? For now, JD’s move lowers the technical bar to using powerful agents, but it also raises fresh questions about who is responsible when an autonomous tool oversteps. It has been reported that the original material was uploaded by a user on Phoenix New Media’s Dafeng Hao platform and that the platform serves only as an information host.

AISpace
View original source →