China warns of 'security backdoor' in Anthropic AI coding tool
3 hours ago
- #AI Tools
- #Cybersecurity
- #China-US Tech
- Chinese regulator warns of a 'security backdoor' in Anthropic's Claude Code AI tool, potentially transmitting sensitive data without consent.
- Anthropic claims the mechanism is an anti-abuse measure to detect fraud and unsupported regions via device timezone and routing checks.
- China's National Vulnerability Database advises users to check, uninstall, or upgrade Claude Code to prevent data leaks, and Alibaba bans its use due to security concerns.
- Anthropic accuses Alibaba of reverse-engineering its AI models, while an engineer states the tracking was an anti-abuse experiment being rolled back.