Broadcom's Symantec Threat Hunter Team reported on 29 October an unauthorised relay service, used to run OpenAI Codex and Claude Code in regions where they are not officially sold, at about a dozen organisations, mostly in mainland China, in sectors such as semiconductors and pharmaceuticals. A setup script rewrote endpoints to relay servers and set Codex's sandbox to "danger-full-access", so that all prompts and secrets transited third-party servers. Symantec says this is not a vulnerability in either product and has not attributed the operator or confirmed active abuse.
From developments to evidence and analysis
Background
No explicit relationship is recorded.
Evidence
No explicit relationship is recorded.
Follow-ups / later versions
No explicit relationship is recorded.
Related questions
Cite this article
AI & Security Observatory（2026-10-29）「Symantec、OpenAIとAnthropicのコーディングエージェント設定を書き換える非公認「AI中継」を報告」『AI & Security Observatory』、第1.0版、最終更新2026-11-01。https://ai-security-observatory.org/ja/brief/brief-20261101-20/
AI & Security Observatory (2026-10-29). Symantec reports an unauthorised "AI relay" that rewrites Codex and Claude Code settings and disables sandboxes at about a dozen organisations, mostly in mainland China . AI & Security Observatory, version 1.0. Updated 2026-11-01. https://ai-security-observatory.org/en/brief/brief-20261101-20/
Version 1.0 · Publication date 29 Oct 2026 · Updated 1 Nov 2026
Revision history
- Earlier revisions have not been recorded.