China's NVDB Labels Claude Code a Security Backdoor

China's government-affiliated National Vulnerability Database (NVDB) formally classified Anthropic's Claude Code as containing a "security backdoor" on July 8, according to an AFP report — reportedly the first time a Chinese government body has labeled a major US AI coding tool a national security threat. Our stance on Claude Code remains conditional: Anthropic confirms the tracking code is being removed and the technical capabilities haven't changed, but the geopolitical risk premium on US AI tools in China-adjacent environments is now explicit and escalating (CBS News / AFP(opens in new tab)).

What happened with Claude Code

According to AFP, the NVDB — affiliated with China's Ministry of Industry and Information Technology (MIIT) — published an advisory on July 8 claiming Claude Code can "transmit sensitive information" including users' locations and identity-related identifiers back to Anthropic's servers without consent.

The advisory instructed Chinese institutions and users to:

• "Conduct a comprehensive check immediately"
• "Promptly uninstall or upgrade to the latest secure version from which the relevant backdoor code has been removed"
• Strengthen network traffic monitoring to prevent unauthorized data leakage

This follows Alibaba's internal ban on Claude Code — effective July 10 — which we covered last week. The distinction matters: Alibaba's ban was a corporate IT policy decision. The NVDB classification is a government-level security action with potential cascading compliance obligations.

Anthropic's response

Anthropic engineer Thariq Shihipar acknowledged the tracking mechanism on X last week: "This is an experiment we launched in March that was meant to prevent account abuse from unauthorized resellers and protect against distillation" (Shihipar on X(opens in new tab)). He said stronger mitigations had been developed and the tracking was being removed — "This should be fully rolled back in tomorrow's release."

Anthropic hasn't responded to AFP's requests for comment on the NVDB classification specifically. The company already blocks users in China and other nations it deems adversarial from accessing its products — though VPN and proxy access remain common.

The escalation chain

Why it matters

The NVDB classification is a political signal, not just a technical one. China's government is now on record calling a US AI coding tool a national security threat — mirroring the US government's own security-based actions against Chinese AI, but applied to developer tooling rather than frontier models.

For enterprises operating across US-China supply chains, the risk calculus on Claude Code just got sharper. You're no longer dealing with a corporate IT policy question — you're dealing with a state-level security classification that could trigger compliance obligations, procurement reviews, and legal exposure for any organization with Chinese operations, partners, or subsidiaries.

The US-China AI tool decoupling is accelerating. What began as export controls on chips and frontier model weights is now reaching the developer tools layer. Claude Code is the first major US coding tool to receive this classification — it will not be the last.

What changes for you

If your enterprise operates in or with China: Add Claude Code to your geopolitical risk register now. An NVDB classification can cascade into procurement restrictions, data-localization requirements, and compliance audits — even after the underlying tracking code is removed. Legal and compliance teams need to treat this as a state-level trigger, not a vendor security bulletin.

Verify your Claude Code version. Anthropic confirms the fingerprinting code is being rolled back. Ensure you're on the latest version without the tracking mechanism. If you're behind a corporate firewall, monitor network traffic for unexpected telemetry to *.anthropic.com endpoints as an additional safeguard.

Watch for reclassification. NVDB advisories can be updated. Monitor whether the classification changes after Anthropic's rollback is confirmed. The political signaling may persist regardless of the technical fix — plan accordingly.

Our Claude Code verdict stays conditional — the tool's agentic coding capabilities remain best-in-class, and Anthropic is addressing the tracking mechanism transparently. But the geopolitical risk premium is now priced in explicitly. For teams in China-adjacent environments, the compliance burden may outweigh the productivity gains.

FAQ

What is the NVDB?

The NVDB is China's National Vulnerability Database, affiliated with the Ministry of Industry and Information Technology (MIIT). It maintains a public registry of software vulnerabilities and security advisories — a classification by the NVDB carries government-level weight.

Does this affect my Claude Code usage?

If your enterprise operates in or with China, yes — the NVDB classification can trigger compliance obligations and procurement reviews. If you operate entirely outside China, the technical impact is minimal: Anthropic confirms the tracking code is being removed. Update to the latest version.