Reverse Engineering Reveals Hidden Code in Claude Code Targeting Chinese Users
aftahi_ai · x · 2026-08-04
A developer reverse-engineered Claude Code and found that Anthropic had hidden detection logic in an obfuscated binary for about three months, without mentioning it in the changelog.
When triggered, the code did the following:
- Checked the system timezone against Asia/Shanghai and Asia/Urumqi.
- Scanned proxies or custom API endpoints against a hard-coded list of about 147 Chinese entities, including Alibaba, Baidu, ByteDance, and DeepSeek.
- Ran multi-factor classification when traffic passed through non-Anthropic endpoints.
Instead of sending results as normal telemetry, the code hid them inside the system prompt using prompt steganography to avoid detection.
More from coding & agent
- Include Standard Libraries in RL Environments to Stop AI Agents Reinventing the Wheel — cephaloform · 2026-08-04
- From GANs to Agent Swarms: An AI Vet Reencounters 'Adversarial Collapse' 12 Years Later — bingxu_ · 2026-08-04
- Building Social Media Agents: API Approvals Are the Real Bottleneck — arslan2012 · 2026-08-04
- Qwen 3.8 Coding Test: Nearly Matches K3 at Half the Price — bindureddy · 2026-08-04
- Overcoming State Loss in Long-Horizon Agents: New Framework Boosts Accuracy — Ziyu Ma · 2026-08-04
- Skip Docker: db-here Enables Zero-Risk Database Isolation for AI Agents — andersonbcdefg · 2026-08-04