OpenClaw faces scrutiny as RCE risk, MIIT curbs surface

What to Know:

• 200,000 OpenClaw instances claim lacks public verification.
• Verified data shows just over 40,000 exposed OpenClaw instances globally.

A claim attributed to the National Cybersecurity Notification Center states there are over 200,000 active OpenClaw instances globally, including about 23,000 in China. That figure is not corroborated by publicly verifiable data.

As reported by Infosecurity Magazine, researchers have documented just over 40,000 OpenClaw exposed instances reachable on the public internet. This verified exposure count is substantially below the 200,000 claim.

Available reporting does not corroborate the specific China total of 23,000. The public evidence supports large-scale exposure, but at a lower magnitude.

Why it matters: remote code execution (RCE) vulnerability and misconfiguration

OpenClaw’s orchestration layer, when publicly reachable, can enable remote code execution (RCE) pathways. Misconfiguration, especially default or missing authentication, turns deployments into accessible attack surfaces.

Practitioners emphasize the real-world risk is exposure and access rather than autonomy. “The immediate risk is not autonomy, but access and exposed infrastructure that attackers can abuse,” said Jeremy Turner, VP of Threat Intelligence & Research at SecurityScorecard.

As reported by Yahoo Finance, China’s Ministry of Industry and Information Technology issued internal notices in early March 2026 limiting OpenClaw installations across government and state-owned enterprises. Those China MIIT restrictions reflect a heightened risk posture toward exposed AI orchestration services.

Active deployments versus publicly exposed: what the numbers mean

Active deployments count installations, including those behind firewalls and on private networks. Publicly exposed counts measure instances reachable from the internet without protective controls.

Risk hinges on exposure. Ten thousand hardened internal deployments may present less systemic risk than a smaller number of internet-facing nodes with weak authentication.

Reconciling the figures, a large “active” number could coexist with a much smaller “exposed” number. Security reviews should prioritize externally reachable interfaces first.