Unauthorised Codex and Claude Code relay service disables sandboxing at 12 organisations

Symantec's Threat Hunter Team reports an unauthorised relay service that deploys legitimate OpenAI Codex and Anthropic Claude Code CLIs and then modifies their settings to turn off sandboxing. It has affected 12 organisations since early 2026, mostly in mainland China. The excerpt gives no CVE IDs, affected versions or indicators of compromise, so no specific remediation is stated. Defenders may wish to review how AI coding agent CLIs are installed and configured in their environments, given that these agents can read files, write code and run commands.

For security analysts · Written by AI from 1 source · Wrong?